Build `read_csv_dict(rows, key_col, val_col)` — the workhorse "turn a CSV into a lookup dict" helper: - `rows` is a list of dicts (think `DictReader` output). - Return a dict mapping `row[key_col]` → `row[val_col]`. - If duplicate keys appear, LAST row wins (mirrors dict-update semantics). - Empty rows → empty dict. - Missing key/val column in any row → skip that row (defensive — partial data common in real CSVs).