fix(alerts-dashboard): consider only alert triggers with anomaly setup

[Samyak2]

This was causing an issue where if anomaly was manually disabled by
setting anomaly_params to NULL, getting the time series frequency
would fail.

[gitpod-io]

[netlify]

✅ Deploy Preview for frontend-sb canceled.

Name	Link
🔨 Latest commit	e138e65
🔍 Latest deploy log	https://app.netlify.com/sites/frontend-sb/deploys/62c2883ac70ac70008b3f80d

[manassolanki]

@Samyak2 I think let's also try to have some provision in the code for disabling/pausing the anomaly for a given KPI after running that for a few weeks without breaking/affecting things elsewhere.
Just give it a thought and let me know if we need to have changes at some other places as well.
cc: @varunp2k

[Samyak2]

@manassolanki I have looked at a uses of anomaly_params throughout the codebase, everything seems fine - the only places where we have used it are either in anomaly, alerts (only in triggers - so it's only run along with anomaly) or anomaly_data_view (which is what sets anomaly_params).

We could add an endpoint to disable anomaly by setting the anomaly_params to None. There is a run_anomaly column in the kpi table, but that isn't being used anywhere. Maybe we should switch to that instead of relying on anomaly_params being None. This will be for v0.10.0, I'm merging this now.