Allow a configurable max timespan per index

[LeeDr]

Describe the feature: This idea came from users at Elastic{ON} 2018.

This user has some indices with few docs (and/or small docs) that they can query over a long timespan in Discover, but then when they switch to a different (much larger) index that same large timespan gets queried and kills their cluster. The idea came up to have a maximum timespan setting per index.

I'm not sure how Discover (or other tabs) would handle the case when the timespan is larger than that maximum setting.

1. Display an friendly error message so the user has to set their timespan to a smaller range?
2. Automatically decrease the start time of the timespan?

This user typically would switch back and forth between indices in Discover but now has to open multiple tabs to use different timespans in each.

[timroes]

We discussed this issue and came to the conclusion of not adding that feature as described here, since we'll think this will be solved by other features.

The main driver behind this is performance optimizations. We are currently in the process of rewriting courier (#34022) which will also allow for better timeout handling (and retries) and query cancellation, which should solve a lot of these issues already.

If you really want to manually limit down the time range of a specific index to not be viewable in Kibana, we recommend you're using filtered index aliases which allow you to create an alias for that index, that filters down to only that time range. That way the limiting already happens on Elasticsearch side and thus be a real limit and not being very easily bypassed, which would happen if we would limit on the Kibana side.