optimize cluster operation and some other fixes

[chris-sun-star]

Summary

1. add task recovery
2. add more integration test for obcluster operation
3. automatically alter sys tenant's locality and zonelist when zone operation
4. check observer indeed exists in obcluster after recover, if not add an add observer process
5. maintain parameters only when obcluster is in running status
6. create service for monitor service discovery
7. fix data race when update status and add a parameter when build docker to enable race detection
8. grant more resource to manager container

Solution Description

[powerfooI]

LGTM