Rotate capped training samples across iterations

backend/tests/test_failure_driven_yolo_sampling.py

@@ -209,6 +209,48 @@ def test_region_cap_drops_only_repeats_and_preserves_every_unique_tile() -> None
     assert metadata["sampled_entries_by_region"]["flanders"] / len(paths) <= .65
 
 
+def test_region_cap_rotates_repeats_between_sampling_rounds() -> None:
+    manifest = {"samples": [
+        {"sample_slug": "fl", "split": "train", "region": "flanders", "context": "industrial"},
+        {"sample_slug": "wa", "split": "train", "region": "wallonia", "context": "rural-town"},
+        {"sample_slug": "br", "split": "train", "region": "brussels", "context": "dense-urban"},
+    ]}
+    summary = {"tiles": [
+        {"sample_slug": "fl", "split": "train", "label_count": 2, "image_path": f"/tmp/fl-{index}.png"}
+        for index in range(4)
+    ] + [
+        {"sample_slug": "wa", "split": "train", "label_count": 2, "image_path": f"/tmp/wa-{index}.png"}
+        for index in range(2)
+    ] + [
+        {"sample_slug": "br", "split": "train", "label_count": 2, "image_path": f"/tmp/br-{index}.png"}
+        for index in range(2)
+    ]}
+    assessment = {
+        "status": "continue_training_loop",
+        "gates": {"min_region_f1": .45, "min_region_precision": .5, "min_region_recall": .4,
+                  "max_pure_empty_false_positives": 0},
+        "calibration": {"regions": {
+            "flanders": {"f1": .2, "precision": .3, "recall": .2},
+            "wallonia": {"f1": .6, "precision": .6, "recall": .6},
+            "brussels": {"f1": .6, "precision": .6, "recall": .6},
+        }},
+    }
+
+    first, first_metadata = MODULE.build_sampling(
+        summary=summary, manifest=manifest, assessment=assessment,
+        positive_repeat=5, max_region_share=.65, sampling_round=1,
+    )
+    second, second_metadata = MODULE.build_sampling(
+        summary=summary, manifest=manifest, assessment=assessment,
+        positive_repeat=5, max_region_share=.65, sampling_round=2,
+    )
+
+    assert first != second
+    assert set(first) == set(second)
+    assert first_metadata["sampling_round"] == 1
+    assert second_metadata["sampling_round"] == 2
+
+
 def test_coastal_precision_failure_targets_port_and_dunes_negatives() -> None:
     manifest = {"samples": [
         {"sample_slug": "coastal-train", "split": "train", "region": "flanders", "context": "coastal-urban"},