surface the new analysis caveats in the workbench

The backend now says when a result covers a different area than was drawn, or
when a score belongs to one confidence cut only. None of that helps an
operator while it stays in the response body.

- the raster selection adapters carry the model-coverage and widened-cell
  warnings into the map panel and mark the result an estimate when either
  applies, so an existing warning slot renders them;
- the map workspace shows the selection-edge disclosure next to the object
  count;
- the detection panel shows average precision and the F1-optimal threshold
  beside the single-threshold figures, and the box-versus-footprint
  interpretation when candidates are detector boxes.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

frontend/src/components/detection/DetectionLab.tsx

@@ -949,6 +949,26 @@ export function DetectionLab({
                 <p>
                   {detectionQaResult.box_to_footprint_diagnostics.possible_box_to_footprint_mismatch_count} mogelijke vormafwijkingen. Precisie, herkenningsgraad en F1 hierboven blijven gebaseerd op de strikte geometrische overlap.
                 </p>
+                {detectionQaResult.box_to_footprint_diagnostics.interpretation ? (
+                  <p>{detectionQaResult.box_to_footprint_diagnostics.interpretation}</p>
+                ) : null}
+              </div>
+            ) : null}
+            {detectionQaResult.precision_recall_curve ? (
+              <div className="detection-qa-diagnostic">
+                <span>Drempelonafhankelijke kwaliteit</span>
+                <strong>
+                  AP {formatNullableNumber(detectionQaResult.precision_recall_curve.average_precision, 3)} · beste F1{' '}
+                  {formatNullableNumber(detectionQaResult.precision_recall_curve.best_f1, 3)}
+                </strong>
+                <p>
+                  {detectionQaResult.precision_recall_curve.best_f1_threshold === null
+                    ? 'Geen kandidaten om een werkpunt uit af te leiden.'
+                    : `De beste F1 ligt bij drempel ${formatNullableNumber(
+                        detectionQaResult.precision_recall_curve.best_f1_threshold,
+                        2,
+                      )}. Precisie en herkenningsgraad hierboven gelden alleen voor de gekozen drempel; AP vat de volledige curve samen en maakt vergelijking tussen modellen mogelijk.`}
+                </p>
               </div>
             ) : null}
           </div>