There are quite a lot of foreachs in ktexteditor, so IMHO it makes sense to split up the porting in multiple commits, so in case of regressions one can bisect in the overall changes a bit and then has fewer changes to reason about where something might have been broken.
So my plan for further commits is:

• one commit where method-local copies would be made explicit ,where currently they are done implicitly in the foreach() by some call -> minimal risk, as not shared
• commits where new structures are generated by iterating shared containers, without method calls which might modify the containers potentially
• commits per classes/submodules where each change needs careful review, as recursive non-const calls happen

Splitting up the changes also makes review easier/more manageable in smaller time units hopefully :)

Also did a review and it looks good to me, thanks!