#7805: Ensure all references to mode are Literals, not str.

[Julian-O]

Made edits, waiting for automated tests to run.

I am unclear how to actually test the changes - e.g. to make sure the example code in #7805 is now passing without type warnings.

[github-actions]

Diff from mypy_primer, showing the effect of this PR on open source code:

vision (https://github.com/pytorch/vision)
+ torchvision/prototype/transforms/functional/_meta.py:156: error: Argument 1 to "convert" of "Image" has incompatible type "str"; expected "Optional[Literal['1', 'CMYK', 'F', 'HSV', 'I', 'L', 'LAB', 'P', 'RGB', 'RGBA', 'RGBX', 'YCbCr']]"  [arg-type]

[srittau]

Thanks! We have a good set of automated tests in typeshed that already test quite a few things. In fact, mypy primer had a complaint above, but in this case I think pytorch should just adapt its type annotations.

We have a few manual tests in test_cases, but for a fairly simple change as this, we don't need any specific tests.

[Julian-O]

Diff from mypy_primer, showing the effect of this PR on open source code:

Hmm... What is going on here? meta_.py makes call to convert (line 156) based on literals listed in line 136-141, one of which is "LA" which is outside the accepted values for _Mode.

Pillow explains that LA is not one of the "standard modes", but that

Pillow also provides limited support for a few additional modes, including:

LA (L with alpha)
[...]

I am not sure what the next step should be,

[hauntsaninja]

Maybe we should add it to _Mode?