DoubleEndedIterator and ExactSizeIterator for To*case.

[clarfonthey]

Because the iterator itself is exact, I see no reason to not implement these traits. Sort of iffy in hindsight about DoubleEndedIterator but I've offered both to see what people think.

[rust-highfive]

r? @brson

(rust_highfive has picked a reviewer for you, use r? to override)

[clarfonthey]

Ping @brson

[alexcrichton]

I'm not sure that this is a guarantee we want to provide perhaps? Is there a motivation for this beyond just completeness?

I could imagine future implementations, for example, that lazily compute future characters as opposed to buffering them all up in memory immediately as soon as the iterator is constructed. Implementing next_back for such an iterator may be much more difficult.

[clarfonthey]

I will agree that shortly after I made this I thought that implementing DEI was a bit unnecessary, although I kept it in anyway. I'll remove that.

ExactSizeIterator seems reasonable, though.

[alexcrichton]

Even that, though, do we get much benefit? It seems like it may preclude "interesting" future implementations maybe

[clarfonthey]

Initially my thought was that it could be used to help figure out more exact allocations for str::to_*case, but in hindsight I'm not sure how useful it would be.

I guess that in principle, it makes sense to know how many characters are required for a lowercase form, but I'm also not sure if that's actually a strong desire.

[alexcrichton]

@brson may have more comments, but I think I personally have yet to be convinced (but certainly could be!)

[alexcrichton]

Ok we discussed this during libs triage today but the general sentiment was along the lines of what I've been saying previously which is that these are niche enough API use cases and without a concrete use case in mind we'd prefer to avoid restricting ourselves.

If you've got a use case in mind for these though feel free to resubmit!