[ML] Single metric model plot info should be shown for partitions that do not contain anomalies

[sophiec20]

Describe the feature:

I have created two anomaly detection jobs jobA:count by host @1m and jobB:count by host @15m as I would like to compare the results from the same analysis using different bucket spans. Both jobs have model plot enabled and the cardinality of host is 4.

Anomalies exist for all hosts in jobA, but only 3 hosts in jobB.

In the single metric viewer for jobA, I can see the model plot for all 4 hosts.

In the single metric viewer for jobB, only host1-3 can be selected from the drop down. The chart label says Single time series analysis of count (4 distinct host values) but I cannot select and view host4.

When checking the .ml-anomalies-* indices, I can see that model plot info has been stored for host4. However because there are no anomalies associated with host4, then I cannot view its model in the single metric viewer.

I would like to be able to view model plot, even if there are no anomalies.

Describe a specific use case for the feature:

This would be useful when debugging why a split field does not contain an anomaly.

[elasticmachine]

Pinging @elastic/ml-ui (:ml)

[peteharverson]

Closing as fixed for 7.6.0 by #53879