groupby() filter performs auto sorting under the hood, and it should support specifying case sensitivity for the sort order, the same way sort() filter does.

Current groupby() implementation always performs case-sensitive sorting under the hood, which makes it inflexible, and behave inconsistently with how the sort() filter works (case-insensitive by default). When used in the same template for the same dataset, groupby() and sort() filters produce different sort order on the same fields.

I think, I would personally prefer to not have any implicit sorting in groupby(), and chain it with sort() explicitly, but I do understand that auto sorting was, probably, added to groupby() for convenience.

Implementing an option to control case sensitivity seems to be pretty straightforward:

• A case_sensitive parameter can be added to do_groupby() function, similar to do_sort().
• make_attrgetter() inside do_groupby() should be called with postprocess=ignore_case if case_sensitive==False.

If this feature request looks reasonable, I should be able to submit a pull request.