[x86] Generate AVX512 fixed-point instructions

[rootjalex]

This PR adds support for generating saturating_(add | sub) and pmulh(rs) on Skylake and Cannonlake (i.e. for AVX512BW). It also increases simd_op_check test coverage of fixed-point operations on those archs.

I also did a bit of clean-up on the way:

• Use LLVM intrinsics for x86 abs
• Remove explicit use of saturating_pmulhrs, do a proper rewrite (as a temporary fix before [x86] Separate vector instruction selection and CodeGen passes #6884 lands).

I did not add abs to codegen because it doesn't appear that LLVM currently exposes non-masked versions of AVX512 abs variants.

Fixes #7002