Azure DevOps - does Push or Pull assume (incorrectly?) a "clean" hosted agent /home/vsts/work folder

[mikewo-dc]

Some behaviours in Azure DevOps Microsoft hosted "ubuntu" agent make me think the hosted agent is not "cleaned" between pipeline runs and AzOps assumes that it will be? NB: For my own experimentation I have AzOps in GitHub and AzOps in Azure DevOps, both identical, pointing to the same tenant etc. The ADO project is a more recent experiment.

Couple examples of weirdness in ADO that I have not seen much like in GitHub yet

• Added a pair of Management Groups (example and example2 under it) - increased from 8 MGs to 10 MGs - run Pull pipeline - Initialize errored out - I could see it counted 8 MGs (and didn't find my new ones) in "Expanding Management Groups"- but then ERROR was trying to retrieve the new MG? see here:
  WARNING: [23:53:21][AzOpsScope] Management Group not found:
  WARNING: [23:53:21][New-AzOpsScope] Failed to: Creating new AzOpsScope object using scope [/providers/Microsoft.Management/managementGroups/example] | Management Group not found:
  WARNING: [23:53:21][Save-AzOpsManagementGroupChildren] Failed to: Creating scope object | Management Group not found:
  Save-AzOpsManagementGroupChildren: Management Group not found:
  ##[error]PowerShell exited with code '1'.
  Seemingly fixed by disabling and enabling the Agent in the Azure Pipelines pool in Organization Settings?

• Deleted a whole LZ Management Group structure - ran Pull process manually in both GitHub and ADO - GitHub recognised the deletion on perhaps the second attempt - but the ADO pipeline continued time and again to try and fail "Expanding Management Groups" for the top level deleted Management Group - several attempts later, Seemingly fixed by disabling and enabling the Agent in the pool?

Is it possible the ADO hosted agent could have stale files the /home/vsts/work folder structure?