Add multi‑commodity support to device scheduler

[Ahmad-Wahid]

This PR adds initial support for multi‑commodity commitments (e.g. gas and electricity) in the device_scheduler, building on the grouped‑devices functionality. Flow commitments can now be tagged with a commodity, and EMS‑level flow constraints are applied per commodity by aggregating the relevant devices.

TODO

• Specify how commodity is encoded per device in the flex‑model
• Exclude non‑matching commodities (e.g. gas devices) from EMS electricity commitments
• Extend flex‑model schema with commodity metadata
• add gas-price field to flex-context
• set up gas commitment using gas-powered devices and gas-price
• Update scheduling.rst
• Add tutorial examples for multi‑commodity scheduling

closes https://github.com/SeitaBV/simulations/issues/987

[read-the-docs-community]

Documentation build overview

📚 flexmeasures | 🛠️ Build #31246494 | 📁 Comparing 1d63893 against latest (36400dd)

🔍 Preview build

Show files changed (13 files in total): 📝 11 modified | ➕ 1 added | ➖ 1 deleted

File	Status
changelog.html	📝 modified
genindex.html	📝 modified
index.html	📝 modified
py-modindex.html	📝 modified
_autosummary/flexmeasures.data.models.planning.html	📝 modified
_autosummary/flexmeasures.data.services.scheduling.html	📝 modified
_autosummary/flexmeasures.utils.html	📝 modified
_autosummary/flexmeasures.utils.job_utils.html	➖ deleted
api/v3_0.html	📝 modified
cli/change_log.html	📝 modified
concepts/commitments.html	➕ added
concepts/device_scheduler.html	📝 modified
plugin/customisation.html	📝 modified

[Ahmad-Wahid]

@Flix6x could you please suggest some assertions for both test cases?

[Flix6x]

Please comment on why the test suite is failing.

[Ahmad-Wahid]

Please try to run these test cases: test_two_flexible_assets_with_commodity and test_mixed_gas_and_electricity_assets

[Flix6x]

I reviewed the two tests.

[Flix6x]

Given that the prices are constant, I'd expect the tie-breaking preference to charge sooner and discharge later would kick in.
The schedules should represent that by showing all charging happens in the first time slots, each up until the max capacity for the device (there is no site capacity).
Please add the assert statements for this behaviour and observe that the schedules are not currently reflecting this behaviour. Therefore, there must be a problem in the code, too. What do the schedules look like when you schedule this system against main (without setting the commodities)? I mean, you'd expect the same result as before, right?

Let's also add a check against the electricity costs for both devices. The expected totals should be explained in the code. For instance, at a 100 EUR/MWh, a total demand of x kWh and an efficiency of y, we expect z costs.

[Flix6x]

What are you testing in this test? Please explain the reason for the test's existence in the docstring.

My take: the gas boiler is currently creating gas, which is pretty weird. :) You'll need a production-capacity set to 0 kW. Afterwards, the gas boiler just stands idle. I would expect this test to have the gas boiler do something, too, and then proof that electricity costs and gas costs are both as expected.
I also think in this test the two devices are essentially completely separate systems. That is, shared commodity, constraints or storage. I think that is worth pointing out.