use 'types-docutils' instead of 'docutils-stubs' for docutils type annotations

[danieleades]

closes #12007

I expect this to be a long-running PR which will help identify gaps in the typeshed docutils libraries.

Once all the issues are squashed upstream this can be merged.

[danieleades]

currently 121 errors: Found 121 errors in 38 files (checked 183 source files)

i'll see what that changes to after python/typeshed#11468 is merged

EDIT now down to 110 errors: Found 110 errors in 33 files (checked 183 source files)
EDIT now down to 97 errors: Found 97 errors in 32 files (checked 183 source files)
Found 85 errors in 29 files (checked 183 source files)
Found 82 errors in 28 files (checked 183 source files)
Found 70 errors in 28 files (checked 183 source files)
Found 59 errors in 26 files (checked 183 source files)

[danieleades]

#12034 should help to reduce the type errors in this PR

[picnixz]

Definitely yes since many of the errors were due to that.

[danieleades]

Found 56 errors in 26 files (checked 183 source files)

getting harder to make progress on this, as a lot of the remaining errors are genuine type safety bugs in sphinx, rather than just gaps in the upstream type stubs.

For example there are many places where Sphinx uses T | None as T and fails to handle the None case. Mypy doesn't like that.

[chrisjsewell]

Cheers for all the work @danieleades!
generally all good, just see the one comment for a quick suggestion

[picnixz]

Haven't looked at everything but I think it should be fine. We'll do PRs to fix each type ignore I think.

[danieleades]

any major blockers to getting this merged?

[picnixz]

I don't think I have something else in mind.

[picnixz]

I'm going to wait until this evening (around 18:00 UTC+1) before merging it to let @chrisjsewell review it again if needed. Otherwise, I'll merge it. Please ping me @danieleades if I forget about it.

[chrisjsewell]

all good cheers!

[picnixz]

Thank you!