Attempt to fix multierror typing

[CoolCat467]

At the moment, this is only ignoring generics not matching properly in the subclass. If anyone has any better ideas on how to handle this, feel free to try it out.

[codecov]

Codecov Report

Merging #2787 (8f2d55f) into master (4d19399) will not change coverage.
The diff coverage is n/a.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #2787   +/-   ##
=======================================
  Coverage   98.94%   98.94%           
=======================================
  Files         113      113           
  Lines       16919    16919           
  Branches     3050     3050           
=======================================
  Hits        16741    16741           
  Misses        123      123           
  Partials       55       55           

Files Changed	Coverage Δ
trio/_core/_multierror.py	100.00% <ø> (ø)

[TeamSpen210]

I think the correct solution is to change MultiError/NonBaseMultiError to also be generic, and propagate that throughout the code.

[A5rocks]

I tried and it was annoying (some branch in my fork...)

[CoolCat467]

I think the correct solution is to change MultiError/NonBaseMultiError to also be generic, and propagate that throughout the code.

That sounds like a good idea, I'll start trying to do that now.

[TeamSpen210]

It probably is fine if we add more ignores here, since it's deprecated anyway. Users should only be annotating/interacting with ExceptionGroup.

[CoolCat467]

It's looking like it might not be possible to make NonBaseMultiError generic, because ExceptionGroup which we are subclassing itself needs to be using the generic we are defining

[CoolCat467]

I've given up attempting to make NonBaseMultiError generic, I don't think it's easily possible without doing some nasty things at runtime with monkeypatching

[A5rocks]

Assuming this passes type checking and that the weird previous changes showing in the diff thing is fixed, this looks good to me.

[A5rocks]

Actually the merge looks quite messed up -- maybe reset to a clean state, rebase onto master, and copy over the type ignore deletion + new type ignore?

[A5rocks]

Looks good