@ocaisa Hmm, I'm not sure this is the best approach, I feel it can lead to continued confusion because we have two variables with different semantics.

Moreover, since self.iter_idx will be >=1 after the first call to handle_iterate_opts, the if self.iter_idx > 0 in prepare_step is wrong (but it doesn't do much harm, it just resets the toolchain also during the first iteration).

Maybe the best way is to have a next_iter method that (currently) simply bumps self.iter_idx (and spits out a log message than the iteration is done)?