The ignore_drop_queries_probability setting (used in stress tests) converts DROP TABLE to TRUNCATE TABLE, but TRUNCATE does not stop the periodic refresh task of a refreshable materialized view. This causes the orphaned view to keep refreshing indefinitely, consuming background pool threads and progressively overwhelming the server — especially under TSan where each operation is 5–15x slower.

Root cause analysis of the "Hung check failed, possible deadlock found" flake in Stress test (arm_tsan):

1. Test 03221_refreshable_matview_progress.sql creates an MV with REFRESH AFTER 10 SECOND
2. The stress test's ignore_drop_queries_probability=0.2 converts the final DROP TABLE to TRUNCATE TABLE
3. The view survives and keeps refreshing every 10 seconds for the rest of the stress test
4. 48 refreshes observed, each getting progressively slower under TSan (from sub-second to 59 minutes)
5. The server becomes completely unresponsive, failing the hung check

Fix: skip the DROP-to-TRUNCATE conversion for refreshable materialized views, since TRUNCATE doesn't stop their periodic refresh task.

CI report: https://s3.amazonaws.com/clickhouse-test-reports/json.html?PR=99997&sha=355f90c5ba8982dc501d353f211c43350cfbc20e&name_0=PR&name_1=Stress%20test%20%28arm_tsan%29
#99997

Relates to #101383

Changelog category (leave one):

• CI Fix or Improvement (changelog entry is not required)

Changelog entry (a user-readable short description of the changes that goes into CHANGELOG.md):

...

Documentation entry for user-facing changes

• Documentation is written (mandatory for new features)

Version info

• Merged into: 26.4.1.676