Add option for backends to run vectorization passes

[gbaraldi]

With this and a change to AMDGPU, for loops in the code can vectorize

[maleadt]

See the TODO below the code you changed:

TODO: Make this not dependent on uses_julia_runtime (likely CPU), but it's own control

The native back-end should then declare can_vectorize = true

Also, doesn't this need to be set for AMDGPU, which seems like the motivation for this change?

[gbaraldi]

Should CUDA also not set this?

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 75.54%. Comparing base (a239ae0) to head (610be9e).
⚠️ Report is 2 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #716      +/-   ##
==========================================
+ Coverage   75.24%   75.54%   +0.30%     
==========================================
  Files          24       24              
  Lines        3635     3639       +4     
==========================================
+ Hits         2735     2749      +14     
+ Misses        900      890      -10     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.