Don't base mechanics on tiers. It will just create more broken multiclass builds. Perfect example is Soulborn vs. Figther: Soulborn gets 8 extra feats over Fighter on this tier basis, with no restrictions on use, alongside their three Incarnum feats. Because you've rated the Soulborn a low tier over its baseline version, it is now strictly superior to Fighter. And because it's feats, it's actually not worth it to take Fighter dips for low-tier classes anymore because multiclassing now takes out your high level bonus feat progression on top of delaying your class features. And, of course, it also ignores how it absolutely buggers multiclass optimization standards, because now Psychic Warrior 7/Soulborn 11/Totemist 2 is a respectable build from Soulborn 12 giving you over a dozen extra feats to be absolutely certain you get everything remotely applicable and plenty enough Essentia for Midnight Augmentation, Girallon Arms, and Dread Carapace.

Point buy being base 25 and differing with LA is a steep issue in that it double-dips penalties hard. A Goliath suddenly finds itself barely able to have an 18 assigned in Strength, having only one other score to give a 10, whereas an Orc will afford 18 Strength, 10 Dex, and 14 Con, giving it extra HP by point-buy on top of having an extra level for that to affect, making the bar for LA+1 to be worth it absolutely insane because it guts your ability to generalize. Many people are going to have a target primary score, come Hell or high water, and all penalizing point-buy is going to do is make them dump anything else harder.

Also, please bring back the lack of stacking crit modifiers, they were removed with the 3.0 changeover for good reason. Thanks to this, a Core-only Falchion or Kukri in the hands of anyone with access to Keen Edge now gets a 9-20 crit range, making it so you are literally incapable of accuracy below 55% without Miss chances added, and the loss of iterative penalties means you're easily getting multiple crits per round. Combine with the above and... Well, Soulborn is now one of the best blunt-force combatants in the game because they have a stupid number of feats to spend on stealing key Soulmelds from Totemist and Incarnate, acquiring Essentia, and obtaining the normal feats for their build of choice.

Spellcasting Monsters should be in the opposite direction because players are not getting ahold of those resources by any sort of shapeshifting, and SLAs don't scale to higher-level SLAs, so for a DM spellcasting monsters should be a valued resource because the existing spellcasting will stack with levels to avoid insane HD bloat and an oversized toolbox trying to give them level-appropriate magic-user capacity for 4+ levels higher. If you insist, I'd suggest moving them to some more-streamlined system with Spells Known, like 5e Pact Magic or a spell-points functionality, and declare their "suggested" list is their required list rather than something flexible so whatever remaining player shenanigans grab it can't repurpose it. If it's about summoning, mandate the specific summon variant (where you choose a single kind, if not exact individual, monster to summon on learning the spell) and throw in proxy slot expenditure, so you're almost certainly disadvantaged over Shadow Illusion versatility.