Fix mishandling of typing.Self in attrs generated inits

[davfsa]

Fix mishandling of typing.Self in attrs generated inits by using the same expansion method as the dataclasses plugin

Fixes#14685

[davfsa]

The test run seems a bit confusing, as locally, it succeeds (outside of the test suite). Is it possible that the test have some configuration that may cause the behaviour to change? (Looking more towards strict_optional)

[A5rocks]

You can specify flags at the top of the test (# flags: --strict-optional might work?)

Additionally I think this expand_type function should go into common.py to reduce duplication

[davfsa]

You can specify flags at the top of the test (# flags: --strict-optional might work?)

The flag didn't seem to do much, specially given that it seems to be the default since mypy 0.600 (as per https://mypy.readthedocs.io/en/stable/kinds_of_types.html#no-strict-optional)

This is really confusing, but I will have another look later

[davfsa]

@A5rocks you were actually right with the flag, i just seemed to have typed flag instead of flags, which is why it didnt work. Thanks for pointing me in the right direction!

I also merged similar code function as you pointed out. Hope it looks good :)

[JukkaL]

Thanks for the PR! A few comments and questions below.

[davfsa]

@JukkaL Addressed one of your points and got questions about the other ones :)

[github-actions]

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