Fix: soc min does not need to be mandatory

[BelhsanHmida]

Description

• Let storage scheduling continue when a flex-model omits soc-min.
• Treat missing soc-min as no global lower state-of-charge boundary, rather than defaulting it to zero.
• Preserve timed soc-minima constraints without filling gaps with an implied global lower bound.
• Clarified the scheduling metadata for optional soc-min.
• Added changelog entry

Look & Feel

N/A. This is scheduler behavior and documentation only.

How to test

pytests:

pytest -k "add_storage_constraints_skips_global_minimum_when_soc_min_is_missing or add_storage_constraints_with_soc_minima_and_missing_soc_min_has_gaps"

Manual check:

• Called add_storage_constraints(..., soc_min=None) directly.
• Confirmed no TypeError is raised.
• Confirmed the generated global min column remains empty.
• Confirmed validate_storage_constraints(...) returns no violations.

Related Items

Closes SeitaBV/ems#211

---

Sign-off

• I agree to contribute to the project under Apache 2 License.
• To the best of my knowledge, the proposed patch is not based on code under GPL or other license that is incompatible with FlexMeasures

[Flix6x]

Nice. I think we may need the same for soc-max. It would be interesting to extend the new tests to a case where soc-min is defined, but soc-max is missing (or both are missing).

[BelhsanHmida]

Nice. I think we may need the same for soc-max. It would be interesting to extend the new tests to a case where soc-min is defined, but soc-max is missing (or both are missing).

Yes, this is now covered. I added tests for soc-min defined with missing soc-max, both bounds missing, and soc-maxima with gaps. I also added preprocessing coverage so missing soc-max no longer raises before constraints are built.

[Flix6x]

I thought about this at length. I now think we should not use ensure_soc_min_max anymore at all. It was only used for the old sensor schedule trigger anyway. I now removed it. This means we no longer try to infer the soc-min or soc-max if they are not passed. The function originally:

1. Tried to get the soc-min/max from the asset. By now this already happens in collect_flex_config, for all fields, so this became redundant.
2. Assumed 0 as the soc-min. Not a good assumption if SoC is used as a proxy for temperature.
3. Tried to get the soc-max from the soc-targets field. Not a good assumption of the soc-max lies below soc-at-start.

The greater contribution of this PR is allowing constraint validation to pass for devices without soc-min or soc-max. That means the StorageScheduler is becoming a more generic DeviceScheduler. I'm not ready to rename it yet, but this is where I feel we are heading.