Filter speakers in Parliament UK corpus

**Is your feature request related to a problem? Please describe.**
The Talking Empire researchers would like a good way to filter for a specific speaker and then search for keywords in the speech content.

This is not possible because there is no filter for speakers. You can search in the speaker field, but then you can't combine that with a different query in the content.

**Describe the solution you'd like**

Add a search filter widget for the speaker field. Trivial to add.

The problem with this is:
- The search filter widget will only show the N most frequent speakers in the results, which may not be reliable.
- Speaker names may not be one-to-one with speaker IDs. (The data does include speaker IDs, but dropdown with ID values would be awful to use since IDs in this corpus are arbitrary number sequences.)

Hence why the filter wasn't added already. We discussed this and want togo for this solution:

Make the multiple choice filter more sophisticated so it can make extra requests to allow searching values (with a backend search).

**Describe alternatives you've considered**
- Advanced query builder: #1436
- Other improvements to rhe search filter:
- virtual scroll to browse through more values
- It can combine a pretty/messy field with an ID field. (I.e. use the ID field for the actual filter, but show the pretty values to the user.)
- Let users activate a speaker (or speaker ID) filter from a document: #729 -> As I mention in that issue, this is pretty easy on a technical level.
- Allow arbitrary values in the "multiple choice filter" so you can type in a specific value that is not in the top N.

Locate the multiple choice filter and the speaker-field search handling, then trace how filter values are requested from the backend. Done means users can search speaker values beyond the most frequent results and combine the selected speaker with a content query.

Written by the indexing model from the issue text.