[fix][meta] Close resources when ZKMetadataStore construction fails - #26673

Fixes #26582

Motivation

When ZKMetadataStore construction fails while connecting to ZooKeeper, superclass executors have already been created. Since the constructor never returns, the caller cannot close the store. The remaining non-daemon threads can prevent initialize-transaction-coordinator-metadata from exiting.

Modifications

• Close superclass resources in the constructor failure path before rethrowing MetadataStoreException.
• Preserve a cleanup failure as a suppressed exception on the original failure.
• Add a regression test that verifies failed construction leaves no metadata-store threads alive.

Verifying this change

• Make sure that the change passes the CI checks.

This change added tests and can be verified as follows:

• Verified ZKMetadataStoreTest.constructorFailureClosesResources fails before the fix because the batch flusher remains alive and passes after the fix.
• Ran the new regression test together with ZKSessionTest.
• Ran ./gradlew --no-daemon --max-workers=1 quickCheck.

Does this pull request potentially affect one of the following parts:

• Dependencies (add or upgrade a dependency)
• The public API
• The schema
• The default values of configurations
• The threading model
• The binary protocol
• The REST endpoints
• The admin CLI options
• The metrics
• Anything that affects deployment

Failed ZKMetadataStore construction now shuts down the metadata-store executors that were created before the ZooKeeper connection attempt.