Add a dedicated metric for corrupted workflows

Is your feature request related to a problem? Please describe.
In our environment when there millions of workflow are running, when there are corrupted workflows due to database layer, it's hard to know which workflows are corrupted.

For the status quo of temporal, corrupted workflows are either:

• Logged & counted in namespace-level metric ( ConditionFailedError , CurrentWorkflowConditionFailedError, WorkflowConditionFailedError)
• Logged only without workflow info (for DataLoss error)

Temporal already have a protection for this case by trimming history branch to last valid node. However, when corruption happens on database side, trimming history node won't work (we have case where workflows are corrupted because of ConditionFailedError, current execution record is gone from C*)
Describe the solution you'd like
Adding a new metrics (eg: workflow_corrupted) with label of workflowID and runID. This is more efficient to detect problematic workflows rather than running a full db scan using something like tdbg.

I can help to come up with a PR if you guys think it's useful.

Describe alternatives you've considered
Parsing log, which i think need more compute power than proactively expose information in Temporal server itself.

Additional context
N/A

Start with common/persistence/execution_manager.go at the linked history-trimming logic, then trace the existing metrics and logging for ConditionFailedError, CurrentWorkflowConditionFailedError, WorkflowConditionFailedError, and DataLoss. Done means corruption cases emit a dedicated workflow_corrupted metric carrying the requested workflow and run identifiers, with coverage for the affected cases.