Salvatore Campagna

I need an approval for this. I didn't manage to use `MultiValueMode` because the change is larger than expected. As a result, I would proceed to merge this and work...

Just a minor comment. For the rest LGTM.

@elasticmachine test this please

I will merge this after the other one...because this will give me some merge conflicts and I prefer to fix merge conflicts here rather than in the other just because...

@elasticmachine update branch

[profiles.zip](https://github.com/elastic/elasticsearch/files/8670456/profiles.zip) The `profiles.zip` archive includes three JFR recordings (each recording is a different Rally race). * `profile-excluding.jfr`: the JFR recording stops the challenge just before running the date histogram operation...

After data shared by @ruflin about usage of field types I am changing the list of fields to be supported adding `annotated-text` to the list of fields to support. This...

The reason why I mentioned the storage aspect of this is that, for `_ignored`, as for other metadata fields, understanding disk usage might also be important to reason about costs...

I was wondering if this can be done at Lucene segment level just by storing an (additional) number, or a counter to be more precise. Thinking about this we have...

@jpountz any idea about this? I think we should not do this unless there is actually a better and more efficient way (with respect to running the aggregation).