[ML] Allow to capture user feedback for anomaly detection jobs results

[Gleb-Gadyatskiy]

We are using Elastic ML to detect anomalies. Our Users receive an email with link to Kibana to see actual anomaly there (ML Job Anomaly Explorer).

We need to give User possibility to provide a feedback for detected anomaly (thumbs up/down plus a text for comments)

Specific scenario example An email "New exception was detected in Application A1" is triggered by Kibana Alerts and sent to an Engineering team. A team member like to click on a link in the email and provide feedback from a dropdown a) True Positive. The team is investigating b) False Positive. Not an actual anomaly, not further investigation required c) Wrong team, the issue is caused by ... assigned to another Team ... for further investigation.

Options 1) Simple: possibility to compose am html link to open Add Annotation popup (we will include this link into Alert email). Should work for both Simple and Classification ML Jobs. 2) Complex: possibility to capture User's feedback and stored it into an index (consider that User has a role allows to Index documents in this index) 3) Ideal: allow to build User entry forms in Kibana and store results in indexes like building Kibana Dashboards

[Gleb-Gadyatskiy]

https://www.elasticaccelerationzone.co/forum/t/need-to-capture-user-input-in-kibana/69 https://discuss.elastic.co/t/need-to-capture-user-input-in-kibana/359387

[elasticmachine]

Pinging @elastic/ml-ui (:ml)