CI failed because rustfmt does not accept trailing spaces.

The question is if we want this guarantee to be library wide or limited to some adaptors like Combinations. I'll let @jswrenn decide on that.

The question is if we want this guarantee to be library wide or limited to some adaptors like Combinations. I'll let @jswrenn decide on that.

Ah yes, I hope I didn't misunderstand the conclusion from the other issue. I originally wanted a guarantee for Combination-like adapters, but I interpreted your comments as saying we don't intend to break iteration order for any adapters.

I could add a blanket "unless otherwise noted" clause, if there are specific adapters we'd like to leave open to further optimizations?

Sorry for the misunderstanding. I don't think we can guarantee order for all adaptors, but I am okay guaranteeing it for the combinations adaptors.

Thank you for the clarification, I've updated this PR so that the guarantee is only shown under Itertools::combinations and Itertools::tuple_combinations. Let me know if all looks good to you.

@jswrenn What do you think of his formulation?

@ryanavella If you could commit jswrenn' suggestions and squash into a single commit, I would gladly merge this.

EDIT: Done.

@jswrenn My approval is not enough to get this merged.