- Automate file creation (so users don't have to hand-make starting .csv file with column names).
- Allow for unique name of file per run (so multiple runs can be saved).
- Figure out if outputting pre-computed result columns is worth it.
- Figure out the sensical bounds for each testing variable (e.g. liquidation incentive should never be larger than 100%).
- Ensure that the TLTV logic can work for all sensible variable bounds, or if not, that the failures do not cause the fuzzing to halt.
- Allow all protocol variables to be wiggle-able. The ones that are currently choosable for the TLTV logic are the ones I identified with python models as being important for the total percentage of money being stolen. I'm not 100% sure that the other variables aren't important.