B6b: fix E1's half-migrated export-scope and UI-action-status callers (run_datasets broken on main)

[trissim]

Main is broken since E1 (#1203): every run_datasets call is rejected, the measured CLI cannot build its run request, and the installed GUI smoke raises. This PR is only that fix and its tests.

What fails on main (a751b0eba, observed)

E1 made the export scope and the UI action status class families and left three callers on the old shape:

Caller	Failure on main
RunDatasets.run passes the request's name member to Session.observation_export_params, which calls scope-class methods	AttributeError: 'ZMQRuntimeObservationExportScope' object has no attribute 'check_export_path'; the operation returns rejected (run_datasets_failed)
openhcs-benchmark run-measured passes the scope class into DatasetRunRequest, whose field is the name view	TypeError: DatasetRunRequest.runtime_observation_export_scope must be ZMQRuntimeObservationExportScope; got AutoRegisterMeta.
scripts/smoke_installed_gui.py reads UiActionInvocationStatus.ACCEPTED	AttributeError: type object 'UiActionInvocationStatus' has no attribute 'ACCEPTED'

On main, tests/integration/test_agent_observation_export_zmq.py, test_session_journey.py and test_measured_cli_session.py give 10 failed, 21 passed (with tests/unit/test_smoke_installed_gui.py); 8 of the 10 are these boundaries. The GUI journey waits out its 300 s stage timeout.

Fix

• RunDatasets.run hands the session request.runtime_observation_export_scope.scope (the class).
• The measured CLI builds the request with RuntimeObservationExportScopeName(args.observation_scope).
• The GUI smoke checks action_result.status is ActionAccepted, in require_ui_action_accepted so a test can call it.

Tests (fail on main, pass here)

• tests/unit/test_run_request_export_scope.py: for every registered scope, RunDatasets.run gives the real Session.observation_export_params the scope class.
• tests/unit/test_benchmark_control.py::test_measured_cli_builds_its_run_request_for_every_observation_scope.
• tests/unit/test_smoke_installed_gui.py::test_gui_smoke_accepts_only_an_accepted_ui_action, over the status family.

Observed: on main's sources 5 failed plus a collection error for the smoke module; with the fix 64 passed. The three integration files with the fix: 3 failed, 12 passed.

Still failing after this PR (not this boundary; fixed in the B6b follow-up)

• test_ordinary_execution_can_export_outcomes_without_value_observation: KeyError: 'well_count' (stale test).
• test_read_only_dataset_initializes_in_a_mirrored_workspace: PipelineOrchestrator has no plate_path (stale test).
• test_stopped_run_is_cancelled_and_refuses_finalization: timing-dependent. A stop that lands during compile-before-execution ends as failed, not cancelled. It passed alone (1 passed) and failed in the full file under load.

B6b · guards 3/3 · guardrails rc=0 · production −12 +21 · tests −2 +158

🤖 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.