Massive Edit, top for visibility: It seems the Oracle is in fact a 4-slot caster per BadLuckGamer, thus resolving the conflict between the repertoire text (3 slots/rank) and the chart (4/rank). This explains exactly where all the power that seemed missing from the shuffling between options went.
(Original Post below, thematic critique still applies but please ignore any comments about the class not getting buffed. It got 4 slots/rank, it got mega-buffed.)
Oracle got a huge rework in PC2, and some typos aside, the idea was executed fairly well! I love the flavor of many of the new feats, and even the ones adapted from old mysteries have been buffed to create something cool. I think it's important, before I get into any of my issues with it, to acknowledge both the creativity that went into it, the execution of a very difficult project, and the creation of effectively a new class in a product that only exists due to WotC messing with the OGL and forcing Paizo to scramble. The time crunch must have been immense, and I don't blame the developers for misfiring a few times in an overall fantastic product.
But part of functionally creating a new class out of an old one is that pain points pop up in the design, and many of those pain points in other classes were addressed with the remaster—not so for this new Oracle. Swashbucklers burdened with needing to juggle a bunch of restricted skills got an auto-scaling skill and better ways to handle finicky Panache issues, Alchemists got a complete teardown to address nonfunctional subclasses without harming the core experience, Barbarian got its penalties to AC removed and its clashing subclasses retuned, Investigator massively improved in usability for GMs and players. And so on.
The thing these things have in common, for the most part, is that the core of the class was preserved. There were components that got tweaked or added to, but even when there were feats changed or otherwise removed, everything felt tied to a unifying vision based on what existed prior. Some people may have issues with individual choices, but I think it's pretty clear that almost all changes were a breath of fresh air to many players in otherwise rickety classes.
The Oracle...not so much.
What changed?
I don't want to dwell on any hypothetical balance issues with the new parts of the class, especially while we have no idea if it's a 3-slot or 4-slot caster. Whether things are too strong or too weak is really a hard line to judge without playing with them! Individual cursebound feats may be too weak or too strong, and, credit where it's due, the inclusion of mystery-granted spells was absolutely a patch on a common pre-remaster point of contention. (Of course, some sets of granted spells are significantly stronger or more thematic than others. Flames is outstanding, for instance; Cosmos is terrible, even including a spell already on the Divine list. But I digress.)
Instead, I'd like to consider the overall class design changes. The mysteries themselves have been gutted entirely, with their curse effects changed and significantly streamlined (for good or ill—more on that later) and all passive and active curse benefits gone. (Oracles also lost a free domain spell.) Oracles now have access to Cursebound feats in their place, feats roughly on par with the general power level of focus spells (some of them literally used to be focus spells!) and functioning almost as a 2nd focus pool with a downside.
Each mystery gets 1 at first level which I see mostly as counterpart to the previous Oracle's innate Domain revelation spell. This is the only one that doesn't cost a feat slot, and it is one of exactly 5 things that the innate class features and mysteries—once build and class defining, with immense flavor if often questionable balance—gets you, the other 4 of which are: 1) the starting set of 3 spells+cantrip, 2) access to a single cursebound feat at 10, 3) the ability to take a domain spell in a specific set of domains, and 4) which spells you can access at 11th level for your repertoire via the now-integrated, non-feat divine access feature, and the same prior set of focus spells.
These cursebound feats are in large part derived from the curse benefits once offered as innate tradeoffs for curse intensity but, crucially, now requiring the use of a feat to obtain for a class which had such features innately.
In sum, mysteries got streamlined, and the revelation focus spell-based curse benefits got turned into a feat-tax-laden quasi focus spell system.
Unlike the other classes people had complaints with that got direct upgrades, Oracle got a sidegrade (assuming for the moment the Oracle is in fact a 3 slot caster) with significant upside for offsetting attrition. It is likely that the class is more powerful now, though all that power comes from feats it didn't have before and therefore didn't have to take before either. All it cost was the unique curse playstyles, their flavor, and a much emptier chassis.
What happened to the curses?
Most mystery curse downsides mostly improved, or were otherwise relatively unaffected. Cosmos actually lost its susceptibility to grabs. Flames lost a lot of damage around it (and self-damage), but only takes a token amount of persistent fire now. Bones has a few harsher penalties but is mostly the same, and Tempest and Battle take a malus to sporadic enemy or player options.
Lore is about as it was. (Namely? Not great.) Life got what I'd say is bit worse, though it's dependent on party comp. And poor Ancestors gets a stacking debuff of the worst kind in the game.
Overall, the balance between mysteries is about as questionable as it was pre-remaster, but they're a great deal less harsh overall for many.
Buuuuuut these tweaks came with a crucial cost: the sometimes playstyle-defining passive benefits of the mysteries were removed entirely!
Battle Oracle no longer can use boosted proficiencies for frontlining. Life Oracle cannot heal better than anyone else. Flames Oracles can no longer dodge their own fire better, Cosmos Oracles lost their damage mitigation, and on and on. These passive bonuses might have been too strong in a world with balanced and less punishing curses, but the existence of unique bonuses at all were vital in making the mysteries feel like significant choices on a ludonarrative level (stealing a term from video game deisgn here!) Choosing a unique set of downsides, 3 spells, and a non-exclusive feat just feels less impactful than picking downsides, upsides, and powerful unique benefits.
Is this entirely a bad thing, though? Well, no. I think the new Oracle is much friendlier to new players, and requires a lot less tinkering to work. There are still balance issues with some of the mysteries, yes, but they are much more straightforward. There are fewer unpleasant surprises nestled in moderate curses and many major curses were just nasty! The shift away from focus spells towards cursebound feats is great for the Oracle's unique interaction with per-encounter resources, offering a less confusing comparison to other classes.
But it's still a big cost, and in a book so dedicated to enhancing the games of those who like and want to play other classes in here, it feels really, really bad that the changes aren't actually aimed at making the existing Oracle class better without killing parts of what made it itself.
In a more glib way, what I'm saying is that, like the pre-remaster classes that this book took the time to improve, the new Oracle almost wants its own remaster in the same way other classes in the book got remastered—to smooth over its lessened class fantasy, to make the changes feel less abrupt, and to try to strike a middle ground between so many of these cool new ideas and the prior class's overwhelmingly strong flavor.
What Would Help?
As I said before, I think there are larger questions about balance between mysteries and the strength of the mysteries (and their newer, shallower benefits) as compared to one another. I'm not equipped to speak to any of that, no matter how much I wish I was. Same for the various errors involved like the mixup with spells/rank or the missing spell DC references for cursebound feats that provoke saves.
What I can say instead is that I have 2 suggestions for how to try to deal with what I see as the two largest stumbling blocks in the new Oracle's construction: weak mystery mechanical flavor and feat taxes to interact with no-longer innate core systems.
First: I think mysteries should get either unique passive benefits—even taking the old set for themselves, if balance permits—and/or freely given feats not shared with other mysteries.
Battle and Life Oracles especially seem like their playstyles have been neutered by the remaster, and I can't imagine that throwing them a bone and reapplying their old modifications in some fashion, weakened or otherwise, could cause issues. Oracle is not a class that's breaking a power ceiling, even with more per-encounter options, so I'm skeptical that this would break the balance.
But if that's too much, simple mystery-unique feats would help give mechanical weight to choices that used to have much more of it.
Second: I think Oracles should get an extra free Cursebound feat choice at a later level.
This is a straight buff, even more than the above, but hear me out: this is exactly how many classes were handled in this book alone! Alchemist and especially Swashbuckler got more free bonuses to their mechanics to make their overall designs operate smoothly, taking what would otherwise be feat taxes and turning them into core bonuses. Think of a Kineticist without free Impulse feats as an extreme counterexample of a class that needs these extra feats to function.
The prior mystery benefits have been moved to feats, but part of this means that you are now functionally asked to take more and more feats to interact with the unique mechanic of your class. I think too many free feats would be a problem, but it would massively soften the blow of current Oracle players feeling like their class features were stripped and they have to pay a feat tax to get any back if there were another way to grab one for free later on.
In Conclusion
Thanks for reading this, and again, I want to state that I don't blame Paizo for what I see as some shaky choices for the Oracle. The whole project was on a time crunch and there were hundreds of things to discuss, change, and fit into a limited set of pages. I think the overall high-level Oracle changes were good ideas, and if there were no prior iteration of the class for comparison, I think people would be much happier about this.
But, well, there is an older iteration. And there are, to the shock of some, people who earnestly enjoyed that older iteration, largely because of its uniquely intense flavor. They're out in the lurch, and I figured I'd try my hand at identifying part of why and some relatively simple ideas on how to claw back what the class lost without making it broken or otherwise excessive.
Oh, and please move already-Divine-accessible Darkness off the Cosmos Oracle granted Spells list and make Lore Oracle's cursebound 4 not kill their ability to function. Those would help too.
tl;dr: Oracle traded flavor for feats and the subclasses lack identity now, it'd be good to have less need to claw back previous benefits through newly extant feat taxes and to see a few more unique options.