Allow passing a callable with type vars in self types

[A5rocks]

Fixes #18400

@ilevkivskyi I'd appreciate if you checked the logic in this PR because I enabled polymorphic inference for binding self and had to handle [T] () -> (T) -> T into () -> [T] (T) -> T... I remember seeing this kind of push down logic before, but I can't remember if there's a common function or if there is an issue with doing that.

[ilevkivskyi]

@A5rocks I will check this one later, sorry (with all the reviews I don't have time/energy to finish my PR adding support for properties with different setter types).

[A5rocks]

This also makes the output on #14322 better, though I think the issue has already been solved:

#------ difference on issues/14322/blocks.json
+ /tmp/tmpcftql_gf:49: note: Revealed type is "def [P] (b_param_1: builtins.str, *P.args, **P.kwargs) -> __main__.B[P`<id>, __main__.ADerived]"
+ Success: no issues found in 1 source file
- /tmp/tmpcftql_gf:47: error: Need type annotation for "BADerived"  [var-annotated]
- /tmp/tmpcftql_gf:49: note: Revealed type is "def (*Any, **Any) -> __main__.B[[a_param_1: builtins.float], __main__.ADerived]"
- Found 1 error in 1 file (checked 1 source file)
#------

[github-actions]

According to mypy_primer, this change doesn't affect type check results on a corpus of open source code. ✅

[A5rocks]

I just ran into this again on another project, it would be nice to have this merged! See python-trio/trio#3260 for that use case.

[A5rocks]

@ilevkivskyi just bumping this!

[ilevkivskyi]

I am keeping postponing this, but unfortunately I still have a lot of work on the whole checkmember.py refactor (it will probably take another 5-6 PRs) plus some fallout from 1.16 release, sorry again.

[CoolCat467]

@ilevkivskyi ping, curious if things are less chaotic now

[ilevkivskyi]

Sorry for delay, it has been a year now, LOL. Unfortunately, a bit busy right now. I will try to go through old PRs at some point soon.

[A5rocks]

Just bumping this another time, just in case! @ilevkivskyi (basically I just want to know the answer to the question in initial PR comment, someone else could review the code...)