Occasionally hitting ResourceExistsError on Helix retry attempts

We are occasionally hitting ResourceExistsError on Helix retry attempts causing those attempts to fail as they are unable to upload their test result blobs. An example of this is the two failing crossgen jobs in this run: https://dev.azure.com/dnceng/internal/_build/results?buildId=2329517&view=results. Looking in the Azure Logs, a blob of the same name was uploaded just a few minutes before, so we need to figure out why those jobs are being run again despite us having a successful upload.