Maybe we need to adjust the rules' perceived maximum password length "of the current format" when JtR knows that something else is stacked on top of the rules. With masks, we can infer the correct new maximum length (subtract the mask's added length from the current format's maximum length), but for other rules stacked on top we should probably just assume a high maximum like our global maximum of 125 for the first ruleset.