BUG: Fix np.abs(nan) returning negative NaN on wasm32/Pyodide (#31433)

This PR fixes a bug where np.signbit(np.abs(np.nan)) incorrectly returns True instead of False.

If  C99 abs doesn't actually flip the sign for some NaNs, then seberg isn't sure we should bother
working around it.  But in practice that seems unlikely.