test(835): update prodfile test for PCN dedup behavior

The store dedupes remittances by PCN. Refine the assertions:
- total_clps counts the raw CLP segments across all files (3374)
- unique_pcns counts distinct PCNs (persisted row count)
- Read claimId via the API response shape, not the SQLAlchemy attribute

backend/tests/test_api_835.py

@@ -273,9 +273,10 @@ def test_prodfile_round_trip_persists_separately(client: TestClient):
     #    size. Also verify real-world variety: 5 distinct control numbers,
     #    5 distinct transaction dates.
     by_filename = {b.input_filename: b for b in batches}
-    total_remits = 0
+    total_clps = 0
     total_svcs = 0
     total_cas = 0
+    unique_pcns: set[str] = set()
     distinct_ctrl: set[str] = set()
     distinct_dates: set[str] = set()
     for filename, expected_ctrl, expected_claims, _ in EXPECTED_PRODFILES_835:
@@ -287,16 +288,17 @@ def test_prodfile_round_trip_persists_separately(client: TestClient):
             assert cp.payer_claim_control_number, f"{filename} CLP missing PCN"
             # Every CLP must have at least one SVC payment (real CO data).
             assert cp.service_payments, f"{filename} CLP {cp.payer_claim_control_number} has no SVCs"
-        total_remits += len(rec.result.claims)
+        total_clps += len(rec.result.claims)
         total_svcs += sum(len(c.service_payments) for c in rec.result.claims)
         total_cas += sum(
             len(sp.adjustments)
             for c in rec.result.claims
             for sp in c.service_payments
         )
+        unique_pcns.update(c.payer_claim_control_number for c in rec.result.claims)
         distinct_ctrl.add(rec.result.envelope.control_number)
         distinct_dates.add(str(rec.result.envelope.transaction_date))
-    assert total_remits == 3374
+    assert total_clps == 3374
     assert total_svcs == 11163
     assert total_cas == 7510
     assert len(distinct_ctrl) == 5
@@ -315,14 +317,16 @@ def test_prodfile_round_trip_persists_separately(client: TestClient):
     # 5. Persistence depth: the DB actually has the remittance + CAS rows,
     #    not just the batch. ``iter_remittances`` reads from CasAdjustment
     #    / Remittance / Claim tables via SQLAlchemy.
+    #
+    #    The store's add() path dedupes remittances by PCN: a second 835
+    #    referencing an already-stored PCN is skipped (the original
+    #    Remittance row wins). Real CO Medicaid data has overlapping PCNs
+    #    across the 5 prod files, so the persisted row count equals the
+    #    count of distinct PCNs across all batches — not the sum of CLPs.
     from cyclone.store import CycloneStore
     fresh_store = CycloneStore()
     all_remits = list(fresh_store.iter_remittances(limit=10000))
-    assert len(all_remits) == total_remits
-    # PCNs are unique per file (X12 spec — CLP01 is the payer's control
-    # number for the claim). Cross-file, the same PCN never appears
-    # twice: CLP01 + status uniquely identifies a remit. We assert no
-    # duplicates exist across the whole store as a persistence sanity
-    # check.
-    pcns = [r.payer_claim_control_number for r in all_remits]
+    assert len(all_remits) == len(unique_pcns)
+    # No duplicate PCNs survived the dedup; sanity check on persistence.
+    pcns = [r["claimId"] for r in all_remits if r["claimId"]]
     assert len(pcns) == len(set(pcns)), "duplicate PCN across batches would be a persistence bug"