On mips, 1**np.nan and np.nan**0 may be NaN not 1

On mips, 1**np.nan and np.nan**0 may be NaN not 1

Done this way not a plain xfail to allow only this difference,
not clearly wrong answers

(The same hardware's "invalid value encountered" warnings,
probably from sNaN/qNaN being reversed, are ignored elsewhere:
d/rules and xfail_tests_nonintel_io.patch)

https://en.wikipedia.org/wiki/NaN#Encoding

Author: Rebecca N. Palmer <rebecca_palmer@zoho.com>
Forwarded: no

Gbp-Pq: Name mips_pow_nan.patch