Docs: mention rounding mode(s)?

[jwdevel]

It would be good to mention what support for different rounding modes the library supports.

One thing that annoyed me about libfixmath is that its docs just says it supports "proper" rounding, but doesn't say what kind (it's actually round-half-away-from-zero). Default for IEEE floating point is "round half to even," (if I recall correctly) though that's more annoying to implement in fixed point.

Some libraries have only one mode supported, some have it configurable; would be nice to know where this one stands.