Support all formatting flags in `std::path::Display`

[alexcrichton]

This regression was initially noted in #43765 which was caused in #42613 where this struct in the standard library no longer respects all formatting flags. This is a tracking issue for supporting all formatting flags all the time, not just when it's conveniently already a &str.

[bstrie]

I'd like to make a test case here but I'm unclear on what the issue is. The PR (#43830) that closed #43765 added tests for the width/fill/alignment formatting flags, and I can't think of what other flags would apply to Path. I'm also not sure what "it" refers to in the sentence "not just when it's conveniently already a &str". However, given that Alex made this issue while discussing the PR, I assume he had something in mind here, so I'm hesitant to close this outright.

[m-ou-se]

Relabeling this as a feature request rather than a regression, considering four years have passed without any comments here requesting for it.

@alexcrichton Can you clarify which flags you are talking about here? See the above comment.

[alexcrichton]

Not that I remember 2017 off the top of my head, but rereading here I believe that the issue is that Paths which are valid utf-8 follow formatting flags correctly, but paths which are not valid utf-8 may not. I suspect at the time I was sure that invalid utf-8 paths did not follow formatting flags, but I'm not sure if that's the case still.