Let retried jobs advance viable/strict (#86821)
Today, even if we retry a failed workflow it succeeds on the retry, viable/strict doesn't advance forward.
Success on retry is proof that the error wasn't with the current commit and that we should in fact promote viable/strict. This PR points to an updated rockset query which will only look at the success status of the most recent job in each workflow
Here's the query edited:
Original query:
https://console.rockset.com/lambdas/details/commons.commit_jobs_batch_query/versions/15aba20837ae9d75?tab=sql
Updated query: https://console.rockset.com/lambdas/details/commons.commit_jobs_batch_query/versions/8003fdfd18b64696?tab=sql
Testing:
Tested the old and new query against commits known to have succeeded on retry
Pull Request resolved: https://github.com/pytorch/pytorch/pull/86821
Approved by: https://github.com/huydhn, https://github.com/malfet
