Revert "[Mono] Add SIMD intrinsic for Vector64/128 comparisons (#65128)"

[simonrozsival]

This PR reverts #65128 in an attempt to fix#66083

[vargaz]

This doesn't seem to solve the problem.

[thaystg]

This didn't cause the problem on CI, I'll close the PR.

[thaystg]

/azp run runtime

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[thaystg]

It looks like that reverting this the timeout issue is gone:

[vargaz]

Note that the lane still takes 2h52m (172m), and the timeout is 3h20m (200m), so even a 10-15% increase in total AOT time would lead to a timeout.

[fanyang-mono]

The step of Send tests to Helix did came down to ~10mins from ~1h. I would consider this PR did resolve the issue. This lane usually takes 2h30mins and more. We could let this PR run another time to get another data point.

[simonrozsival]

@fanyang-mono I'll re-run the tests

[simonrozsival]

The llvm aot lanes are now OK:

but so are the same lanes in the rolling build on main:

Now it's Mono minijit Pri0 Runtime Tests Run OSX x64 release and Mono monointerpreter Pri0 Runtime Tests Run OSX x64 release that are timing out (both this PR and the rolling build).

[fanyang-mono]

Since the llvmaot and llvmfullaot lanes are back to normal on main. This PR should be closed. The long Send tests to Helix step that I observed before could be infrastructure hiccups.