HDDS-16629. Allow same-rack placement when no other rack has a writable node

[peterxcli]

What changes were proposed in this pull request?

This builds on #11403 (HDDS-16628) and includes its commits. To see only what this PR adds, use this comparison.

Take a cluster with two racks and decommission one whole rack. Every container with a replica on that rack needs a new copy, and SCMContainerPlacementRackAware wants that copy on a different rack from the existing replicas. The only other rack is the one going away, so placement fails, the containers stay under-replicated, and the decommission never finishes.

This PR makes two changes:

• If no usable node turns up on the other racks, placement now falls back to any rack (when fallback is enabled). It already did this when the other racks had no candidates at all; now it also does it when the candidates it tried were all unusable.
• MisReplicationHandler now checks whether copying the container to the chosen node would actually improve its placement. A copy to a rack that already has a replica doesn't: the container becomes over-replicated, the extra copy is deleted, it is mis-replicated again, and the cycle starts over. In that case the handler now skips the copy.

We found this with the SCM simulation (HDDS-16627).

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-16629

How was this patch tested?

• TestSCMContainerPlacementRackAware: when the other rack is being decommissioned, the new replica goes to the same rack, and without fallback, placement fails.
• TestRatisMisReplicationHandler: no copy is sent when it would not improve the placement.
• In the SCM simulation from HDDS-16627. Add a deterministic simulation test for SCM #11349, with this PR and HDDS-16628. RackAware placement should try other replicas' racks before giving up #11403, 100 seeds over 2, 3, 4 and 12 racks all pass. Without them, 6 of those seeds leave a decommission stuck.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.