Fix hardware optimization compilation - [MOD-6567]

[GuyAv46]

Describe the changes in the pull request

We previously discovered that the way we add architecture optimization flags was wrong, and we "leaked" higher optimizations onto lower optimization level functions (e.g. an SSE function was compiled with the -mavx flag, and yielded a binary that an SSE only machine cannot use).

from GCC documentation:

GCC depresses SSEx instructions when -mavx is used. Instead, it generates new AVX instructions or AVX equivalence for all SSEx instructions when needed.

These options enable GCC to use these extended instructions in generated code, even without -mfpmath=sse. Applications that perform run-time CPU detection must compile separate files for each supported architecture, using the appropriate flags. In particular, the file containing the CPU detection code should be compiled without these options.

This PR fixes the ill-formed CMakeList.txt files and adds new .cpp files for choosing the optimization function, which will each be compiled with the right flags

Note that a regression is expected for the SSE functions, as they got "too optimized"

Mark if applicable

• This PR introduces API changes
• This PR introduces serialization changes

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 95.41%. Comparing base (0fca479) to head (764ffcc).

❗ Current head 764ffcc differs from pull request most recent head 02f3f07. Consider uploading reports for the commit 02f3f07 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #438      +/-   ##
==========================================
+ Coverage   95.39%   95.41%   +0.01%     
==========================================
  Files          66       69       +3     
  Lines        4148     4184      +36     
==========================================
+ Hits         3957     3992      +35     
- Misses        191      192       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[meiravgri]

Great!
Consider to add this to the description:
https://gcc.gnu.org/onlinedocs/gcc/x86-Options.html

GCC depresses SSEx instructions when -mavx is used. Instead, it generates new AVX instructions or AVX equivalence for all SSEx instructions when needed.

These options enable GCC to use these extended instructions in generated code, even without -mfpmath=sse. Applications that perform run-time CPU detection must compile separate files for each supported architecture, using the appropriate flags. In particular, the file containing the CPU detection code should be compiled without these options.

[alonre24]

All good! just simplify a bit benchmarks where possible

[github-actions]

Backport failed for 0.6, because it was unable to cherry-pick the commit(s).

Please cherry-pick the changes locally and resolve any conflicts.

git fetch origin 0.6
git worktree add -d .worktree/backport-438-to-0.6 origin/0.6
cd .worktree/backport-438-to-0.6
git switch --create backport-438-to-0.6
git cherry-pick -x 7e2adad94e7f24f04740bc6482d14abba07cae7c fd6b8478146602f6e4bd323c2133b567da72008f bf1f0240e88b4bc554790f1ecec2783317eef4f1 b1cce60c8585604dce9761832b94c8cc0ef8a227 ebab36dd0198d2bb494462779d8f764b992caa5a 89f8eb08e273807fa97897872d38eb008b55c5f7 b2b87e768e1da1926ca1bafd60a100e8049c4e9b 764ffccbd156e7ce6bb55a76fe8ae4541c7d5370 02f3f07298fe90312f072e131efabaa85b5d7ccb

[github-actions]

Successfully created backport PR for 0.7:

• [0.7] Fix hardware optimization compilation - [MOD-6567] #439