Fix csv columns potentially being numbered wrongly in the header when exporting

[iamllama]

Originally reported by @brishtibheja in https://forums.ankiweb.net/t/bug-guid-column-is-marked-as-deck-column/53924

With the way csv columns are currently numbered when exporting, a column selection that isn't a contiguous subsequence of guid, notetype, deck, tags* causes them to be numbered wrongly in the csv output.

By contiguous, e.g.
Fine: guid, notetype, deck or notetype, deck or deck etc.
Not fine: guid, deck or notetype, tags or guid, notetype, tags etc.

*the field(s) before tags don't affect this

[dae]

I think I'm missing something - why fall back on notetype/guid here when deck_column() does that already?

[iamllama]

It's for when the deck column isn't exported, in which case deck_column returns None and tags_column then has to fall back to the next earliest column(s)

Otherwise, the tags column would be 1 + 0 + the number of fields, potentially overlapping with the guid and notetype columns if they exist

This is the same for all the columns, they have to keep falling back to the next earliest non-None column

[dae]

🤦 thanks :-)

[dae]

@RumovZ if you have any preferences, please let us know

[RumovZ]

Could have been handled better when I originally wrote it, so we're not reliant on getting the column order right in all the various places. But I think it's reasonable to just fix the current logic.