mandatory and non-mandatory scheduling models need to implement time window management as well, so we're going to revise them. This means we'll be redirecting some of the at-work subtour resources to fixing the existing man and non-man models.
Continued clean-up of implementation
pipeline arguments implemented and tested now
testing on windows fixed
testing can't run boxed on Windows, and so data leaked between tests
orca injectables makes all this difficult to manage, so we wrapped orca with an inject wrapper class which allows us to track all orca actions and then added a "roll all inputs back to their initial state at the start of the tests" function
we're slowly moving away from the orca injectables and plan to eventually just load data explicitly when we need it