VMSS VM is in Deleting state, Owner Machine of AzureMachinePoolMachine is not deleted

[mweibel]

/kind bug

What steps did you take and what happened:
I noticed a few Machines with Failed state recently:

$MachinePoolName-42             $Cluster   $MachinePoolName000016           azure:///subscriptions/$Subscription/resourceGroups/$RG/providers/Microsoft.Compute/virtualMachineScaleSets/$MachinePoolName/virtualMachines/42          Failed    12d

The bug seems to be as follows:

1. Scale Set VM is Deleting is logged - VMSS VM is in phase Deleting
2. This marks the AzureMachinePoolMachine as deleted (setting DeletionTimestamp)
3. AMPM.reconcileDelete then deletes VMSS VM
4. once VMSS VM is gone, the AzureMachinePoolMachine gets completely removed
5. Machine stays in Failed state

The behavior contract between Machine and InfraMachine doesn't specify how this should be done, I believe. However, given the current behavior, I believe step #2 is wrong and instead of marking the AzureMachinePoolMachine as deleted, it should instead mark the Machine (i.e. the Owner) as deleted.

What did you expect to happen:
Machine gets deleted successfully and doesn't stay in cluster.

Environment:

• cluster-api-provider-azure version: 1.15.2 with PR 4928 applied
• Kubernetes version: (use kubectl version): 1.28.5
• OS (e.g. from /etc/os-release): Linux