Tag deletion not propagated to all Raft peers, causing data inconsistency

Description:

When deleting a tag in a Raft-enabled Orchestrator cluster, the deletion is not propagated to all Raft peers.
However, when adding a tag, the change is propagated correctly across peers.
This leads to data inconsistency in tags between Orchestrator nodes.

Details:

• Adding a tag: Works correctly because the data is published to all Raft peers via a PublishCommand call.

Relevant code:
https://github.com/percona/orchestrator/blob/bd65e056a73002bcfeae403ac5f3768571877311/go/http/api.go#L1975

• Removing a tag: Does not publish the change to Raft peers.

Relevant code:
https://github.com/percona/orchestrator/blob/bd65e056a73002bcfeae403ac5f3768571877311/go/http/api.go#L1989..L2008

Proposed fix:

The existing Raft command applier logic (already present for tag addition) should also be invoked in the untag method, ensuring the deletion is propagated cluster-wide.
https://github.com/percona/orchestrator/blob/bd65e056a73002bcfeae403ac5f3768571877311/go/logic/command_applier.go#L80..L83

Impact:

Without this change, tag deletions only apply locally on the node where the request was made, leaving other Raft peers with stale tag data.

Start in go/http/api.go by comparing the tag-addition path around lines 1975 and the untag path around lines 1989-2008, then read the existing tag command handling in go/logic/command_applier.go around lines 80-83. Done means tag deletion invokes the existing Raft applier and the deletion reaches every Raft peer without leaving stale tag data.