Migrate MacOs wheel binary builds to ephemeral M1 runners (#110432)
Surprisingly there are no speed difference between running the cross-compilation on `macos12-xl` (x86_64 12 core machine) and `macos-13-xlarge` (m1 6 core machine)
Most of the changes are on the https://github.com/pytorch/builder side:
- https://github.com/pytorch/builder/commit/50a6e91f972158529a7c9f06c4c6fd4510196e32 skips installing mkl on M1 machines
- https://github.com/pytorch/builder/commit/bbb29b0467ecbf4fa1f14e51f30f68a0d7c9dc23 same for llvm-9
- https://github.com/pytorch/builder/commit/8bcc83dbb1e947f7e4c546ea19959d775e7fa7d6 bumps minimal numpy version to 1.19 (as 1.17 is not available for m1)
- https://github.com/pytorch/builder/commit/cc4f1f9055b9e32eb4c81d8525f10a537c913274 skips building tests/distributed for M1
Pull Request resolved: https://github.com/pytorch/pytorch/pull/110432
Approved by: https://github.com/kit1980