[OpenStack] Fix: Retry admin privilege GRANT on RDBMS creation to avoid Error 1044

[powerkimhub]

• Replace the single 15s-timeout attempt in grantAdminPrivileges with up to 5 retries using exponential backoff (5s/10s/20s/40s, 2min overall deadline)
• Only retry on transient network errors (net.Error Timeout()/Temporary()); non-network failures (e.g. bad credentials) still fail fast with no added delay
• Fixes ERROR 1044 (42000): Access denied for user 'admin'@'%' caused by the DB instance's port not yet being reachable right after Trove reports ACTIVE, which previously made the one-shot GRANT silently fail (logged as a warning only) while CreateRDBMS still returned success
• Follows the same retry pattern already used for IBM/Tencent transient network errors ([Docs] Fix: Regenerate stale Swagger docs for NodeGroup APIs #1871/f45231e7)