Fix --limit subject filter dropped for chunked tables in sqlite import.py

[Chessing234]

Bug

mimic-iv/buildmimic/sqlite/import.py's --limit N option is silently ignored for any table large enough to trigger chunked reading (chartevents, labevents, emar, …). The resulting SQLite database contains every row of those tables, defeating the purpose of --limit.

https://github.com/MIT-LCP/mimic-code/blob/5706978/mimic-iv/buildmimic/sqlite/import.py#L160-L170

if os.path.getsize(f) < THRESHOLD_SIZE:
    df = pd.read_csv(f, dtype=mimic_dtypes)
    df = process_dataframe(df, subjects=subjects)
    df.to_sql(tablename, connection, index=False)
    row_counts[tablename] += len(df)
else:
    # If the file is too large, let's do the work in chunks
    for chunk in pd.read_csv(f, chunksize=CHUNKSIZE, low_memory=False, dtype=mimic_dtypes):
        chunk = process_dataframe(chunk)
        chunk.to_sql(tablename, connection, if_exists="append", index=False)

The small-file branch passes subjects=subjects to process_dataframe, but the chunked branch calls process_dataframe(chunk). process_dataframe's subjects parameter defaults to None, and its df.loc[df['subject_id'].isin(subjects)] filter is a no-op when subjects is None.

Root cause

Missing subjects=subjects argument in the chunked call, so chunked tables skip the subject_id filter entirely.

Fix

Pass subjects=subjects in the chunked branch, matching the non-chunked branch so the --limit N filter is applied consistently.