Handle namespaceNotActive error in parentClosePolicy workflow

[yycptt]

What changed?

• Handle namespaceNotActive error in parentClosePolicy workflow
• If chlid domain active in a different cluster, signal the system workflow in the remote cluster to apply the parent close policy.
• We can't rely on frontend autoforwarding for Terminate/RequestCancel, because there're some request fields we need to use that only exist in history service interface.

Why?

• Currently if child domain performs a failover after parent close policy receive the signal, the signal request can not be processed.

How did you test it?

• Added unit test

Potential risks

• if remote cluster is down then, the workflow will keep retrying sending the signal, but this is the same as today because we will keep retrying on namespace not active error.

Is hotfix candidate?
no

[yux0]

what about add cluster name in the identity?

[yycptt]

Good idea 👍 . Let me check how to get current cluster name here.

[yux0]

If it is going to touch many places, I am fine without the cluster name. I just feel it may help with debugging.

[yycptt]

it's a simply change. I have added current cluster name to processor's params and used it in the identity.