Add deployment_status metric

[isum]

This PR adds a new deployment_status metric with the following behavior:

• Once graph-node has figured out that it should index a deployment, deployment_status is set to 1 (starting);
• When the block stream is created and blocks are ready to be processed, deployment_status is set to 2 (running);
• When a deployment is unassigned, deployment_status is set to 3 (stopped);
• If a temporary or permanent failure occurs, deployment_status is set to 4 (failed);
  • If indexing manages to recover from a temporary failure, the deployment_status is set back to 2 (running);

No other values for deployment_status should be expected.

Closes #5405

Breaking changes

The deployment_failed metric is removed and the failures are reported by the new deployment_status metric.

Node operators that relied on deployment_failed should switch to using deployment_status.

Testing

The new metric was tested on a local graph-node and Prometheus setup. I have tested various scenarios such as errors, pauses, restarts and the metric behaves as expected and reports correct values.