feat(investigations): add event impact and work-order feedback flows

Add isolated burst and pollution impact evidence, parallel work-order workflows, persisted notifications and graph updates.

Checkpoint current code. Known follow-ups remain for large pollution result previews and graceful geometry failure.

tests/unit/test_scada_presentation_coverage.py

@@ -85,6 +85,25 @@ async def investigation():
     return task, repository, service, evidence.evidence_id
 
 
+@pytest.mark.anyio
+async def test_pollution_charts_validate_quality_values_and_keep_their_own_workflow():
+    from app.services.scada_presentation_coverage import add_targets, applies, ScadaObservations, verify_chart
+    task, repository, _, evidence_id = await investigation()
+    task["workflow"] = "water_pollution"
+    evidence = repository.evidence[evidence_id]
+    for device in evidence["data_snapshot"]["data"]["data"]["devices"]:
+        device.update(device_type="quality", measurement_unit="mg/L")
+    value = chart(devices=[DEVICES[0]])
+    value["scada"].update(metric="quality", unit="mg/L")
+    assert applies(task)
+    verified = verify_chart(task, value, [evidence], [evidence_id])
+    assert verified["metric"] == "quality" and verified["gaps"] == []
+    observations = ScadaObservations(task, [evidence])
+    assert add_targets(None, [DEVICES[0]], observations)["targets"] == [DEVICES[0]]
+    task["handling_context"] = {"phase": "field_review"}
+    assert not applies(task)
+
+
 def coverage(comparison_devices):
     slots = [(START + timedelta(minutes=15 * i)).isoformat() for i in range(13)]
     common = {"coverage": {device: {"covered": slots} for device in DEVICES}, "metric": "flow", "window_start": START.isoformat(), "window_end": END.isoformat(), "gaps": []}