i386 lib/msun: SSE2 is always present on amd64 host

[markj]

lib/msun/i387/fenv.c

70

This is still referenced in lib/msun/i387/Symbol.map. So presumably the symbol must be kept, and the implementation can always return 1?

[kib]

Slap one more HAS_SSE()
Restore test_sse and __has_sse.

This passed buildworld.

[markj]

lib/msun/i387/fenv.c

39

__SSE_YES is 0, so this return value is inverted.

[kib]

Return '1' from __test_sse().