[MusicXML import] guess page one credit words type

In MuseScore 3.5, all MusicXML page one credit-word elements are imported into a vertical frame as formatted untyped text. As typically no text type information is present in the MusicXML file, previous MuseScore versions tried to guess the text type based on position and font size. Unfortunately, this is quite unreliable. Nevertheless, at least one complaint has been received about removing this feature (https://musescore.org/en/node/309655).

My personal opinion is that detecting the title is probably worthwhile (as it is most likely simply the entered text using the largest font size), but detecting composer versus lyricist and copyright versus subtitle seems too unreliable to be worthwhile. Whether my assumptions are correct or incorrect depends on how scores are most typically formatted, which I do not know.

Options:
1: leave as is in 3.5
2: try to detect the title only
3: restore the pre-3.5 behaviour
4: others ?

Opinions are welcome.