[trissim]

K1 · PR #1210 · guards 8/8 · production −1513 +1706 · tests −1693 +1898

Receipt: docs/refactor/consolidation-20261009/K1-plane-modes.md. Most of K1 landed through G3; this PR does the remainder.

What changed

• One plane-axis family. RuntimePlaneAxis is an AutoRegisterMeta family (RuntimeSlicePlaneAxis, SourceBindingPlaneAxis, NoPlaneAxis); the enum-keyed strategy family and ImagePayloadMetadataCompositionMode (which restated it one to one) are merged into it and deleted.
• Typed absence. plane_axis is always a member (NoPlaneAxis replaces None). A projection is a WholeAxisProjection or a SelectedPlaneProjection; RuntimePlaneProjection is a WholeStackProjection or a StackPlaneSelection. Only the selecting members have plane_index.
• ObjectLabelDomainScope is a family that owns the behaviour of the former ObjectLabelPlaneDomainStrategy family.
• Payload methods. image_metadata, array_data, scalar_source_provenance() and slice_source_provenances() are runtime-value family methods; image_metadata_of / array_data_of are deleted. G3's projection gate passes.
• B5's G3 rows. Volume projection: three were a production bug (an absent scalar source record became an empty one in with_source_context_from), fixed; the fourth pinned behaviour Selected-plane artifact finalization fails with viewer streaming and no storage axes #395 changed on purpose and now asserts Selected-plane artifact finalization fails with viewer streaming and no storage axes #395's. Derived role: both were stale tests (the test bypassed the runtime's declared TIFF write).
• E1's guard table loses the three enums' 36 rows.

Production adds 193 net lines: the family members, the four projection classes and their docstrings replace enum members and None checks; about 600 of the churn is the scope strategy classes moving above the dataclass that uses them.

Deletion certificates

• derived: ImagePayloadMetadataCompositionMode (equalled RuntimePlaneAxis through for_plane_axis; the axis family now owns composition); RuntimePlaneAxisStrategy and leaves, ObjectLabelPlaneDomainStrategy (methods now on the family members); image_metadata_of / array_data_of (family methods); the "must be payload, plane, or None" TypeError in SourceImageObjectLabelBuildRequest.plane_semantics and the "requires a runtime-slice, source-binding, or channel axis" check in project_declared_source (the family has no other leading member; require_member validates at construction); the ObjectLabelValue type check moved from object_measurement_runtime_inputs to prepare_object_labels.
• contract (runtime formats reset): plane_axis is spelled "none" instead of null in viewer image metadata, debug facts and CellProfiler profile fields.
• tests-only: none.

Checks run locally

• 30-workflow CellProfiler parity under the shared flock: 1 passed in 712.88s at 1f57f180a.
• Structural ratchet (openhcs, scripts, benchmark): exit 0, empty delta. nra-r1: exit 0, "increased": [] (run with a 600 s budget; the machine was loaded).
• 118 touched test files: 5 failed, 3960 passed, 4 skipped. The five: two integration tests that also fail on main (test_bbbc021_canonical_illum..., test_neurite_outgrowth_runs_on_synthetic_plate_as_2d_channel_stack); the zmq witness, which connects to other agents' execution servers on this machine and passes when run alone; two test_cellprofiler_library_loading tests that fail only in that combined run (a module registered by test_classification_interop_boundaries leaks across files) and pass alone; I did not run that combination on main.

🤖 Generated with Claude Code

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.