math:fmod: Implement a floating-point remainder function

[mansourmoufid]

Hello,

This implements a floating-point remainder function, similar to the integer modulo operation.

This function can be useful for trigonometric operations which expect input to be bounded, for example to [0, 2pi].

It is done in three parts. First, the _fmod function is defined as a static inline function in a header file, so that its code can be re-used internally by other developers (of sine, cosine, etc). Second, the p_fmod_f32 function is declared as part of the library. Finally, the function p_fmod_2pi_f32 may be useful since [0, 2pi] is the typical domain of the trigonometric functions.

I implemented these as an experiment after discussion in #166. I hope others can find them useful.

[mansourmoufid]

By the way, I wasn't sure how to generate the gold data file, so I just used a bit of Python:

>>> import math
>>> import random
>>> n = 100
>>> x = [random.uniform(-100.0, 100.0) for i in range(n)]
>>> twopi = 2.0 * math.pi
>>> _fmod = lambda x: math.fmod(a, twopi) + (0.0 if x > 0.0 else twopi)
>>> data = [(a, 0.0, 0.0, _fmod(a)) for a in x]
>>> for (a, b, c, d) in data:
...     print("{:.6f},{:.6f},{:.6f},{:.6f}".format(a, b, c, d))

Let me know if that's incorrect.