Document determinism guarantee of FxHasher

[benediktsatalia]

Currently, FxHasher is deterministic (accross multiple instances within one run, accross runs, accross machines with the same target_pointer_width). I am wondering if this crate is committed to upholding this rule or if this cannot relied upon with future changes?

I am asking because I think determinism is an important part for many applications (think about when you want you iterate over a FxHashMap and want your runs to be reproducible). So if the crate is committed to uphold this rule it would be great if this could be documented somewhere.

[WaffleLapkin]

I wouldn't recommend iterating over a hashmap like ever. Even if the order is determenistic, it's not stable -- little input changes can drastically change it.

But either way I think this is a reasonable guarantee to provide.