Supplemental support for numerical and mathematical functions

In response to #179, I checked the support status and performance of some other functions. The (incomplete) list is shown below.

To be supported

[ ] ^(::FixedPoint, ::Integer) for more accuracy
- e.g. N0f8(0.8)^3 != N0f8(64/125)
[x] bitstring (PR #195)
[x] big (Issue #184, PR #186)
[x] clamp (Issue #179, PR #194)
[x] copysign(x::Fixed, y) and flipsign(x::Fixed, y) for speed performance (PR #187)
[ ] fma and muladd
[x] rationalize (Issue #184, PR #186)
[x] 3-arg rem and div (cld) (PR #226, PR #230)
[ ] rem2pi and mod2pi
[x] signbit for speed performance (for Normed) (PR #187)

[x] abs(::Normed) specialization is not helpful. (PR #187)
- The fallback implementation works fine.
[x] copysign(x::Normed, y) and flipsign(x::Normed, y) should be throw error. (PR #187)

maxintfloat (maxintfixed and minintfixed?)
parse and tryparse for fixed point numbers with suffix
bit shift operators (Issue #250)
- I think they are usuful but somewhat dangerous.
numerator and denominator
widemul