Problem

start_cluster() calls stages sequentially but has no health verification between them. If MySQL servers are started but not fully healthy (due to BF-1 / #34), start_mysql_mgmt() proceeds and fails when MySQL Router can't connect. The root cause is obscured by downstream failures.

File: src/initialize.py, start_cluster()

Fix

def start_cluster(self) -> None:
    self.start_mysql_servers()
    self._verify_mysql_cluster_ready()  # New: confirm all 3 nodes healthy
    self.start_mysql_mgmt(MASTER)
    self._verify_mgmt_ready(MASTER)     # New: confirm Router + keepalived
    self.start_mysql_mgmt(BACKUP)
    self._verify_mgmt_ready(BACKUP)
    self.start_superset()

Acceptance Criteria

• Explicit health verification added between each deployment stage in start_cluster()
• _verify_mysql_cluster_ready() confirms all 3 MySQL nodes healthy before start_mysql_mgmt()
• _verify_mgmt_ready() confirms Router and keepalived healthy before next stage
• Failure at a stage produces a clear error pointing to that stage, not a downstream symptom