Description of problem:

CAPOA controlplane does not handle metal3machine cleanup:

• when race conditions triggers, and more than necessary machines are being created, the associated metal3machine is created but machine creation fails. This lead to orphhaned metal3machines

  Version-Release number of selected component (if applicable):

How reproducible:

This reproduces only when the right race condition is activated: it went undetected with upstream CAPM3 1.11 and 1.12, but it has been triggered with CAPM3 downstream 4.21
It's not happening 100% of the times, but often enough (around 50% with my setup). When this doesn't happen it seems it's because a lot of worker nodes are being spawned by core CAPI

Steps to Reproduce:

1. Provision a cluster with MCE 2.11 (CAPI+CAPOA+CAPM3)
2. observe "machine storm"
3. ...

Actual results:

• more machines than requested are being spawned

  Expected results:

• only requested amount of machines being spawned

  Additional info: