Fix exponent wrapping in Math.frexp(BigFloat) for very large values

[HertzDevil]

BigFloats represent their base-256 ** sizeof(LibGMP::MpLimb) exponent with a LibGMP::MpExp field, but LibGMP.mpf_get_d_2exp only returns the base-2 exponent as a LibC::Long, so values outside (2.0.to_big_f ** -0x80000001)...(2.0.to_big_f ** 0x7FFFFFFF) lead to an exponent overflow on Windows or 32-bit platforms:

require "big"

Math.frexp(2.0.to_big_f ** 0xFFFFFFF5)  # => {1.55164027193164307015e+1292913986, -10}
Math.frexp(2.0.to_big_f ** -0xFFFFFFF4) # => {1.61119819150333097422e-1292913987, 13}
Math.frexp(2.0.to_big_f ** 0x7FFFFFFF)  # raises OverflowError

This PR fixes it by computing the exponent ourselves.

[straight-shoota]

question: How is this removal related to the other changes?

[HertzDevil]

It too uses LibGMP.mpf_get_d_2exp, so it suffers from the same issue as Math.frexp

[straight-shoota]

I see. And with its removal, calls fall back to reduce_num(value : Float) which uses Math.frexp.