So in Variant A, "[, ...]" is intended to apply to the immediately preceding token but in variant B it is intended to apply to all preceding tokens in the same group.
But (TEXT, TEXT) is valid.
The reality is that there is a trade-off between explicitness and readability (human usability). Complaints strictly about inconsistency are not going to be acted upon. Specific documentation will be improved upon given sufficient demonstration of a usability problem AND the available of a more usable alternative. That the wrong choice will simply "fail fast" in cases where the capturing of the repetition is ambiguous, and there are examples to make clear which is correct, causes one to err on the side of readability for the synopsis (which is unique to each command) rather than consistency between commands.
