Data structure improvements about ckeditor5_sections

[pmelab]

On output filtering: A lot of filters would just work on JSON (e.g. token replacement), but everything that expects valid HTML will break. Technically we could just run JSON -> HTML, apply output filters and then extract data again. But this would mean we lose the performance benefit of already storing parsed data.
Another option could be tagging output filters as compatible and only allow these.

[dan2k3k4]

We should also keep in mind how translations would work, sending a translation job via TMGMT - would we send the JSON to be translated or the computed HTML?

I suppose the diff module would use the computed HTML.

[pmelab]

I think TMGMT is save. When using it "on site" it would just invoke the configured field widget, and for translation services we need special treatment anyway, like we have it on daimler now.

[pmelab]

Another thought on output filtering: I think the "cleanest" way would be to do the filtering on-demand per property in GraphQL or Twig, since this would be conceptually the same as what paragraphs would do with output filtered fields. But I'm not sure if this would be feasible or not.