Optimize `group_round_message`

[mmaker]

Use Iterator::step_by in https://github.com/mmaker/tinybear/blob/main/src/sumcheck.rs#L48 instead of trivially copying indices