Implement LowerExp and UpperExp for integers

[clarfonthey]

I'd be willing to work on this at some point, but I figured that I'd offer this as a feature request in case someone else is interested.

It's very useful to be able to print large integers in scientific notation to get a glance at what order of magnitude they are. For example, seeing 1e10 is much easier to comprehend than 10000000000.

[phungleson]

Thanks @clarcharr, looks interesting and simple enough for me.

I can look into this.

[phungleson]

@clarcharr seems like the implementation for f32 and f64 is around this function

rust/src/libcore/fmt/mod.rs

Line 1539 in 86d9ed6

fn float_to_exponential_common<T>(fmt: &mut Formatter, num: &T, upper: bool) -> Result

Do you think we can just reuse it?

[hanna-kruppe]

That code is strongly tied to floats. I don't believe it can be reused for (64- and 128-bit) integers at all, but even if, it would probably be much more painful than just writing a clean-room implementation tailored to integers.

[dtolnay]

I agree that a separate implementation for integers would be easier than trying to use the float one. I would love a PR to add these impls!

[maxbla]

I fixed this with #66721, but this issue wasn't automatically closed