Inspecting v3.0.2 code i found that BtModuleTextModel render text for verses in parallel mode for each module separately, without passing context of all rendering modules. That makes impossible to translate versifications.
Inspecting rendering pipeline i found correct that text items are selectable for each module separately, so requirement to render
a single verse in parallel view looks justified.
I also would note that i do not get yet, why that model use multiple data roles rather than utilize columns.

So i would make following suggestions to resolve this issue:

1. Add additional parameter to render only single module in CEntryDisplay::textKeyRendering() and still pass full module list.
2. Convert key parameter of CEntryDisplay::textKeyRendering() to sword::SWKey or somewhat like that so it would be casted to VerseKey and translated to v11n of requested module.
3. I think it also would be possible to still render entry for all parallel modules in base text model and add proxy model that will split text in a way suitable for rendering in QML view. Maybe it will require some extra work, like caching data, and extra markup in html but it will improve base text model.
4. V11n translation would be done before calling rendering, in model. This way we will have duplicate code.

In sense of performance i think 1. and 2. are better, but i still would like to have flexible basic text model with good design so it would be easily reused and improved.