Should throw more explicit error message when cluster_manager config is not correct.

currently, if we run a tiflash without the correct cluster_manager config, it won't report but just failed in replica sync.

we should let tiflash throw more explicit error message when cluster_manager config is not correct.

mysql> select * from information_schema.tiflash_replica;
+--------------+------------+----------+---------------+-----------------+-----------+----------+
| TABLE_SCHEMA | TABLE_NAME | TABLE_ID | REPLICA_COUNT | LOCATION_LABELS | AVAILABLE | PROGRESS |
+--------------+------------+----------+---------------+-----------------+-----------+----------+
| test         | s          |       53 |             1 |                 |         0 |        0 |
+--------------+------------+----------+---------------+-----------------+-----------+----------+
1 row in set (0.00 sec)

Start by tracing how the cluster_manager config is parsed and how replica sync reports failures; the issue provides no source file or test path. Reproduce the shown information_schema.tiflash_replica query with an incorrect configuration, then verify that TiFlash reports an explicit configuration error instead of only showing an unavailable replica.