feat(normalizer): add _deduplicate() to persons_tree

tools/import-normalizer/persons_tree.py

@@ -173,3 +173,37 @@ def _parse_row(row_num: int, fields: dict) -> dict:
         "_spouse_raw": spouse,
         "_bemerkung_raw": bemerkung_out,
     }
+
+
+def _deduplicate(persons: list[dict]) -> tuple[list[dict], list[str]]:
+    """Remove duplicate rows. Two-stage:
+
+    1. Exact (firstName, lastName, birthYear) match.
+    2. (firstName, lastName) where the later entry has birthYear=None and an earlier
+       entry already has a known birthYear.
+    """
+    seen_full: dict[tuple, str] = {}   # (first, last, year) -> rowId
+    seen_name: dict[tuple, str] = {}   # (first, last) -> rowId of first entry with a year
+    result: list[dict] = []
+    skipped: list[str] = []
+
+    for p in persons:
+        first, last, year = p["firstName"], p["lastName"], p["birthYear"]
+        key_full = (first, last, year)
+        key_name = (first, last)
+
+        if key_full in seen_full:
+            skipped.append(f"{p['rowId']} duplicates {seen_full[key_full]} ({first} {last}, year={year})")
+            continue
+
+        if year is None and key_name in seen_name:
+            skipped.append(f"{p['rowId']} duplicates {seen_name[key_name]} ({first} {last}, no birth year)")
+            continue
+
+        seen_full[key_full] = p["rowId"]
+        if year is not None:
+            seen_name[key_name] = p["rowId"]
+
+        result.append(p)
+
+    return result, skipped