Use SSE builds for Python Build Standalone on linux

[konstin]

Currently, we're using x86_64-unknown-linux-gnu-install_only_stripped build for linux, but there are builds for x86_64-v2, x86_64-v3 and x86_64-v4 that use SSE to speed up Python (https://gregoryszorc.com/docs/python-build-standalone/main/running.html). We should switch to at least x86_64_v2 (SSE4), maybe even x86_64_v3 (AVX2). x86_64_v4 use AVX-512 which is not available on end user machines, only on server xeons.

[zanieb]

I could have sworn there was a request for this already, but don't know where. Regardless, yes we should support these.

[notatallshaw]

use AVX-512 which is not available on end user machines, only on server xeons.

All new AMD CPUs support AVX-512, both server and consumer: https://en.wikipedia.org/wiki/Advanced_Vector_Extensions#AVX-512_CPU_compatibility_table

Intel consumer CPUs are more complex, some older consumer CPUs do support AVX-512, but newer ones don't.