Tag flow-path limit timeslices with their region's demand condition

[nick-gorman]

Background

Two tables on this branch describe transmission limits by demand condition, and they were inconsistent. The custom-constraints templater (#112) emits region-tagged timeslices read from PLEXOS — qld_peak_demand, vic_winter_reference, and so on. The flow-path limits templater, sourced from the IASR workbook, emitted region-agnostic ones — plain peak_demand, summer_typical, winter_reference. AEMO defines these demand conditions per region, so the flow-path limits should carry the same region-specific timeslices as the rest of the model. For an intraregional link the region is implied, but for an interregional link a region-agnostic timeslice is also ambiguous: a NSW↔QLD limit has a peak_demand value in each direction with nothing to say which region's demand condition each belongs to.

PLEXOS tags a path's forward limit with the destination region's demand condition and its reverse limit with the origin region's (the receiving region's load is what tightens an import limit). The investigation is written up in #109. This PR adopts the same convention for the flow-path limits so they're consistent with the custom constraints: each limit's timeslice is prefixed with the lowercased region id, and both tables end up using one vocabulary.

What to expect in the diff

The logic is one new helper, _add_region_to_timeslices, slotted into the existing orchestrator. Most of the line count is the test files relearning the prefixed values and the docstring I/O examples being brought up to date.

A few design choices worth knowing before reading the code:

• Prefixing happens at the sub-regional level, before any granularity aggregation. A REZ connects its REZ id (geo_from) to its parent sub-region (geo_to), both in the same region, so its symmetric limit should get the same prefix in both directions. single_region later rewrites geo_to to the single NEM geo — so prefixing afterwards would tag the forward limit nem_* and the reverse qld_*. Prefixing first gives both qld_*, and aggregation only re-keys the path id, leaving the prefix alone.
• The region lookup comes from sub_regional_geography, which already maps both sub-regions and REZs to their region id. NEM regions are added as identities so already-aggregated geos still resolve.
• A missing geo fails loudly. _assert_geos_have_regions runs before prefixing for every granularity, so an endpoint absent from the geography raises rather than silently producing a NaN prefix.

Where the changes live

src/ispypsa/templater/
└── transmission.py                          ← all the logic: _add_region_to_timeslices,
                                                _build_geo_region_lookup, _assert_geos_have_regions
tests/test_templater/
├── test_transmission.py                     ← sub_regions cases + helper unit tests
├── test_transmission_nem_regions.py         ← nem_regions cases
├── test_transmission_single_region.py       ← single_region cases
└── test_create_ispypsa_inputs_template.py   ← end-to-end wiring

The timeslice column isn't consumed downstream yet, so this is a no-op at runtime; it brings the flow-path timeslices into line with the custom constraints for when both are used together.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Files with missing lines	Coverage Δ
src/ispypsa/templater/transmission.py	98.83% <100.00%> (+0.12%)	⬆️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.