Use fabsf() for single-precision absolute value

[c0d1f1ed]

Overloads of fabs() for both double and float are only available from C++'s header, not the C header.

Every other math function already correctly used the non-overloaded function name; this was the only outlier. While NanoVDB is written in C++, this change helps avoid requiring C++ headers to be included, which significantly affect compilation time, especially for projects performing just-in-time compilation at runtime.

[linux-foundation-easycla[bot]]

The committers listed above are authorized under a signed CLA.

• :white_check_mark: login: c0d1f1ed / name: Nicolas Capens (664ff55dcf4b792d7ced1924b220043c57488b01)

[kmuseth]

Closing this PR since I'll adopt it elsewhere (see review comment for an explanation)