Term frequency visualisation is unclear about search fields

Currently, the term frequency visualisation will:

- ignore everything except text content fields
- ignore stemming

However, this is not communicated to the user. I had actually forgotten about the first point myself.

Re. that choice (excluding metadata fields, even if they are included in the query), this was implemented in part because the People & Parliament group requested it, but there were also technical reasons (https://github.com/CentreForDigitalHumanities/I-analyzer/pull/761).

The second point is just a shortcoming, I'm not sure if it's a regression or we just never got around to it.

In any case, this could be addressed by adding a "count frequencies in: X" element like the neighbouring words visualisation has.

Start at the term frequency visualisation and compare its menu with the neighbouring words visualisation, which already shows a “count frequencies in” element. Done means the term frequency UI clearly communicates that it uses text content fields and does not apply stemming, using the proposed wording or an equivalent.

Written by the indexing model from the issue text.