it's because the nvidia drivers are opaque blobs, rather than a source distribution.
If nvidia distributed their drivers as open source, i would imagine developers would likely "fix" their games properly because they'd be able to see what is going on underneath the hood, and write more optimal code.
of course, this removes some "competitive advantage" nvidia has over their AMD counterpart.
all of what you say could apply to AMD or Intel but doesn't loo
[0] https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/src/uti...
[1] https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/src/uti...
Yes, these should be fixed properly in the programs instead of the driver but more often than not they won't users don't care that its the program being buggy when it works with another driver because it either doesn't trigger the bug or has its own workaround. The opposite also exists of course - application workaround for driver bugs. Or for bugs in crappy third-party code that gamers like to inject into the process, e.g. the Steam overlay.
If anything, AMD drivers probably have more workarounds as devs are more likely to test with NVidia due to the unbalanced market share.
https://techreport.com/review/how-atis-drivers-optimize-quak...
Personally, I don't believe this is a good example; as this was ATI purposefully crippling the experience (downscaling textures, being the prime aspect) to cheat benchmarks. OP is more referring to "transparent" optimizations (the same experience, but made to run faster for the GPUs architecture, usually via operation reordering or shader modifications).
[0] https://registry.khronos.org/vulkan/specs/1.3-extensions/man...
Most likely when a game is released, the team that build it disbands to work ons omething else and wont have time for performance hacks like this.