HDDS-16631. Fix chooseRandom for rack names with common prefixes

[adwetojha]

What changes were proposed in this pull request?

InnerNodeImpl#getLeaf used a raw string-prefix comparison to determine
whether an excluded scope belonged to a child subtree. As a result, an
excluded path under /r10 was also treated as belonging under /r1,
reducing both subtrees' available leaf counts and causing chooseRandom
to return null even when an eligible node existed.

The change uses Node#isAncestor(String) for topology-aware path
comparison. This preserves exact-node and genuine-descendant exclusions
while respecting path-component boundaries.

A regression test creates nodes under /r1 and /r10, excludes the node
under /r10, and verifies that the node under /r1 is selected.

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-16631

How was this patch tested?

• Focused regression test:
  mvn -ntp -pl :hdds-server-framework test '-Dtest=TestNetworkTopologyImpl#testChooseRandomWithExcludedScopePrefix' -DskipShade -DskipRecon -DskipDocs
• Full affected module:
  mvn -ntp -pl :hdds-server-framework clean test -DskipShade -DskipRecon -DskipDocs
  (646 tests, 0 failures, 0 errors, 1 skipped)
• Checkstyle:
  ./hadoop-ozone/dev-support/checks/checkstyle.sh
• Fork build-branch workflow:
  https://github.com/adwetojha/ozone/actions/runs/37128788136