manage multiple waste outputs according to chemical compatibility

[Poofjunior]

This PR:

• makes it possible to have multiple waste vessels, each of which can be marked with what chemicals it can accept.
• tracks the contents of the reaction vessel during pauses so that it knows where to dump the waste when it resumes.
• adds more validation checks to ensure a Job can run on a machine that can handle the waste.

Heads-up! The bulk of the changes are in brainwasher.py and job.py. (You can probably ignore the bulk of the *.yaml file changes.

More detailed changes:

• move starting_solution out of the job history field so we can record the solution of the reaction vessel during a pause.
• make Vessel, ReactionVessel, and WasteVessel dataclasses. (Also consolidate them)
  • WasteVessels accept a set of all the chemicals you can fill them with.
• add a get_compatible_waste_vessel_id method
  • by passing in a set of chemicals, this method returns the waste vessel index that can accept that kind of waste. If both are acceptable, it returns the vessel with less waste.
• drafted tests for the Job pydantic model.

Reviewing

Heads up! If you're curious about liquid-handling functions like prime_reservoir_line, drain_vessel, or run_wash_step, let's do a demo or walk through the code together! The next best thing is to reference this P&ID diagram.

Testing:

• running jobs works in simulation.
  • pausing jobs correctly populates the reaction vessels current solution
• running a short job works on the real machine. ("chemicals" are just colored liquids.)

TODO

• Test all the changes on real hardware using colored dyes.
• Test that purge can purge to different waste vessels or to the reaction vessel.
• ensure the waste vessel solutions increase each time we dump to them
• ensure the reaction vessel solution contents is being tracked correctly
• reconstitute starting state knowing the contents of all vessels in the event of a power failure or crash. <-- punt this for later after we wrap up chemistry tests. (Kevin and I can always recover the machine from a power failure / crash without additional features. (See: Save/Reconstitute Starting State #24 for requirements.)