Add proper long double is double double support to fpbits

[michaelrj-google]

See: #113235

PowerPC uses double double as its long double type. We should add at least basic FPBits support for it. This will be complicated since the fractional part of the float isn't contiguous.

[llvmbot]

@llvm/issue-subscribers-libc

Author: Michael Jones (michaelrj-google)

See: https://github.com//pull/113235

PowerPC uses double double as its long double type. We should add at least basic FPBits support for it. This will be complicated since the fractional part of the float isn't contiguous.

[michaelrj-google]

Need to change it to 106, since that's what clang/ppc uses for ldbl mant bits