Exclude VMs in specific states from cluster workload optimization by DRS #8073

Description
Right now DRS takes into account for cluster workload optimization even the VMs in UNDEPLOYED and STOPPED states. But it shouldn't.
Although the following messages appears in /var/log/one/oned.log file:

Tue Sep  8 10:13:00 2026 [Z0][PLM][I]: [migrate] Migrate action is not available for state UNDEPLOYED
Tue Sep  8 10:13:00 2026 [Z0][PLM][I]: [migrate] Migrate action is not available for state STOPPED

To Reproduce

1. Put some VMs into UNDEPLOYED and/or STOPPED state.
2. Generate optimization plan.
3. Execute it.

Expected behavior
DRS shouldn't take into account for cluster workflow optimization at least VMs at these states.

Details

• Affected Component: DRS
• Hypervisor: N/A
• Version: 7.4.0 and most likely master

Additional context

Note 1: Maybe the issue should be rather about defining how we want DRS to consider the VMs in the states other than RUNNING or PENDING. We should also decide if this should be handled by DRS or outside of it.

Note 2: it seems that creating datastore requirements, which are passed as the input to DRS Parser, is problematic for the VMs that are not in the running or pending states. Namely, a user observed the case where the last datastore from the history, which is potentially a good candidate, wasn't listed in the requirements at all.

Progress Status

• Code committed
• Testing - QA
• Documentation (Release notes - resolved issues, compatibility, known issues)