[stdlib] Refactor uses of `StaticTuple` to use `InlineArray` about mojo

[vidur2]

Do you have any idea how to run the testing suite? Is the 'Test Examples' workflow comprehensive?

[artemiogr97]

@JoeLoser there are some differences between StaticTuple and InlineArray so I could not replace it in most of the places:

• In os._linux_aarch64, os._linux_x86, and os._macos StaticTuple is used for the last unused member of _c_stat which is register_passable and so all members must be register passable, but InlineArray is not register passable (not sure if we can add the decorator)
• in collections.vector, utils.inlined_string StaticTuple is instantiated with uninitialized values which is forbidden for InlinedArray, maybe a staticmethod undefined_init could be created for those cases? I think keeping the init that causes a compile time error makes sense, should we allow to create an uninitialized array somehow something like __init__(*, uninitialized: Bool) such that it can not easily be created by accident
• the other uses are either in StaticIntTuple or in the tests, so to be removed at the end once they are no longer used

[ConnorGray]

Hi @artemiogr97, re. your second question:

in collections.vector, utils.inlined_string StaticTuple is instantiated with uninitialized values which is forbidden for InlinedArray

Yes, it would be good for us to preserve a way to create uninitialized InlinedArray values.

We may want to go a different direction down the line, but I think your suggestion for a __init__(*, uninitialized: Bool) seems like a reasonable solution to try for the time being🙂

[JoeLoser]

Fixed with 6af1fa7. Thank you!