coverage fails at high freq

Fails on
2023-07-31T09:40:37.5940558Z 69 - sql_cmd_test (Failed)
2023-07-31T09:40:37.5940959Z 88 - sql_cluster_availability_test (Failed)
1. catch failed status, the log is hard to debug now
sql_cmd_test DBSDK/DBSDKTest.LongWindowsCleanup/1 said demo1 deployment exists in tablet, but drop succ in the last time. Check about tablet sp_cache_
cicd workflow ~40m 39s may be this failure
https://github.com/4paradigm/OpenMLDB/actions/runs/5722199470/job/15504895984

The issue mentions sql_cmd_test and sql_cluster_availability_test failures in CI. Start by examining the test logs from the provided GitHub Actions link to understand the error about 'tablet sp_cache_'. Look at the test files in the repository to see how coverage is measured and where the failure occurs. Determine if the issue is in the test setup, the database's cache mechanism, or the CI environment itself.

索引モデルが issue の本文から書いたものです。