SchemaStore shutdown may exceed the graceful timeout during snapshot initialization

[lidezhu]

What did you do?

CI failure: https://prow.tidb.net/jenkins/job/pingcap/job/ticdc/job/pull_cdc_storage_integration_heavy_next_gen_legacy_safepoint/340/stages/

During shutdown, SchemaStore.Close may wait for RegisterKeyspace to release keyspaceLocker while it is initializing or retrying a schema snapshot. The initialization was not canceled by the SchemaStore lifecycle, so shutdown could exceed the 30-second graceful timeout.

The encrypted snapshot path also used context.Background() and log.Fatal, preventing cancellation and potentially terminating the process on normal shutdown.

What did you expect to see?

• Cancel in-progress keyspace initialization before waiting for keyspaceLocker.
• Reject keyspace registration after SchemaStore.Close.
• Propagate cancellation through schema/table encryption.
• Return encryption and cancellation errors instead of calling log.Fatal.

What did you see instead?

As described above

Versions of the cluster

Upstream TiDB cluster version (execute SELECT tidb_version(); in a MySQL client):

(paste TiDB cluster version here)

Upstream TiKV version (execute tikv-server --version):

(paste TiKV version here)

TiCDC version (execute cdc version):

(paste TiCDC version here)