Temporary interface: Add fsm_determinise_with_config.

[silentbicycle]

This interface will probably need some discussion / changes before it's ready to upstream. I think it's worth having some kind of limit here, since determinisation is most likely to have wide variation in resource usage.

• The names (particularly the enum) are cumbersome here, but I would really rather have a distinct return value for exceeding the passed-in limit, rather than using something that nominally fits something from errno, like E2BIG.
• It seems a bit silly to pass in a struct with just the one field, but as long as they have a reasonable default when zeroed, we could add other fields to the struct later without needing interface changes. We could just give fsm_determinise an extra step_limit argument, but if we're potentially going to add any other options at any point (logging flags?) then a struct would be less hassle.
• I'm adding a separate function fsm_determinise_with_config to avoid an interface change with fsm_determinise directly (which is called all over the place). We could either change fsm_determinise (once) and take a struct, come up with a naming convention for <function>_with_an_extra_configuration_struct (fsm_determinise_with_config? fsm_determinise_cfg?) and keep both (and possibly add similar variants for other functions later), or just add a state_limit param, but that would be my last choice.

[katef]

This would be fine to go upstream

[silentbicycle]

Closing, posted to upstream instead: katef#498