Zero shadow price for over-supplied commodities

I've noticed in the `two_outputs` model that a couple of commodities have a price of zero for some timeslices (actually `-0.0`, to be precise):
- `DIESEL` has a price of zero in winter 2020
- `GASPRD` has a price of zero in summer 2020

Both of these are produced by assets with more than one output commodity, and in both cases the price of zero coincides with the commodity being over-supplied across these seasons, which is necessary for the asset to meet demand of the other commodity it outputs.

In this model it seems to only happen in the base year (perhaps a sign of poor calibration), but I see no reason why it couldn't happen in later years where MUSE is choosing the investment decisions.

I see this was predicted as a potential issue [here](https://github.com/EnergySystemsModellingLab/MUSE2/pull/806#issuecomment-3273869038), but don't think this has been followed up on.
@ahawkes Do you have any ideas how we might address this?

I've also come across zero prices while working on #986, where hydrogen appears free, which is at least one reason you can end up in an infinite loop. That might be unrelated though