Write annotations feature requests

[dale-wahl]

I noticed some things that could be improved with the write annotations processor that could be improved when fixing it to work with ndjson files.

1. New filtered dataset: Explorer view does not include parent annotations
   • If you run write_annotations and create a new filtered dataset, it does not inherit the actual annotations. The annotations are added to the dataset as new fields, but not displayed in Explorer with some dataset types (those that do not include all fields in the style sheets).
2. New filtered dataset: Preview and CSV download does not include new annotation fields for ndjson filetypes
   • the fields are added to the ndjson file itself, but due to map_item not including the annotation field names are not shown in preview or in the normal CSV download (the NDJSON to CSV processor works still)
3. Explorer allows annotation field names that could already be field names
   • As a separate database table, this is likely not an issue, but when running write annotations, it can and will overwrite data.
   • This is compounded with ndjson, since a user is not likely to be fully aware of all fieldnames (only the map_item fieldnames).

[sal-uva]

All great points 👍 Will try to get to this soon(ish).

[sal-uva]

Point 1 is now "fixed" since parent datasets are updated, instead of creating a new one. Point 2 and 3 are fixed as of e839a4cf

There's some leftover issues:

• Removed annotation fields are still included in the root dataset instead of removed (maybe this is okay?)
• Annotation fields are not included in processors. This would involve more advanced edits to map_item, some of which is already facilitated by this pull request: https://github.com/digitalmethodsinitiative/4cat/pull/365