slight update for compress by day

src/adsb/combine_chunks_to_csv.py

@@ -126,7 +126,8 @@ def download_and_merge_base_release(compressed_df: pl.DataFrame) -> tuple[pl.Dat
     # Reorder columns to match
     compressed_df = compressed_df.select(base_df.columns)
     
-    # Concat and deduplicate by icao (keep new data - it comes last)
+    # Concat base (old days) with new data (new days)
+    # No deduplication needed since they represent different UTC days
     combined = pl.concat([base_df, compressed_df])
     print(f"After concat: {len(combined)} records")
 

src/adsb/compress_adsb_to_aircraft_data.py

@@ -123,6 +123,10 @@ def compress_multi_icao_df(df: pl.DataFrame, verbose: bool = True) -> pl.DataFra
         # partition_by with as_dict=True returns tuple keys: (date, icao)
         date_val, icao = group_key
         compressed = compress_df_polars(group_df, str(icao))
+        # Set time to start of UTC day for consistent deduplication
+        compressed = compressed.with_columns(
+            pl.lit(date_val).cast(pl.Date).cast(pl.Datetime).alias('time')
+        )
         compressed_dfs.append(compressed)
     
     if compressed_dfs: