Implement time variable MarketBidCost for thermals (only incremental)

[GabrielKS]

In this sequel to #1284, I add the remaining MarketBidCost time variability for thermal generators: the "min gen cost" OnVariable coefficient and piecewise incremental variable costs. Extension to non-thermal generators, and implementation of time variable decremental costs for loads, are left to a future PR, as is the writing of greater-than-two-dimensional parameters to the results (#1304).

[GabrielKS]

Draft for now as I have only handled the min gen cost so far.

[GabrielKS]

Progress update: I've successfully validated time-varying min gen cost, breakpoints, and slopes for incremental thermal MarketBidCost!

Remaining TODOs in this PR:

• Solve bug Bug: MarketBidCost power constraint does not account for min gen power #1318 for real. There is a stopgap in there right now, but it's not great
• Verify the rest of the formulation, e.g., should the generator's active power limits play a role at all?
• Rebase (or more realistically at this point, merge) to eliminate merge conflicts with psy5
• Some code cleanup in the implementation. I don't think we need that bespoke subtype of FunctionData after all, there's some incremental vs. decremental stuff that could be done more cleanly, some lingering TODOs, etc.
• A few more tests that I doubt will catch anything huge
• Some test refactoring and cleanup

Things I propose to do in subsequent PR(s):

• Handle the non-thermal case
• Handle the decremental case
• Write 3D parameters to results (currently blocked by Support greater-than-two-dimensional results outputs #1304 @daniel-thom)

Miscellaneous notes:

• It seems like sometimes, rather than reporting an infeasibility, HiGHS will solve a relaxation of the problem and silently give you that. For instance, I was getting fractional OnVariable values. Might be worth digging into and reproducing more minimally