Apply subjects filter to chunked reads in sqlite/import.py

[Chessing234]

Bug

mimic-iv/buildmimic/sqlite/import.py provides a --limit N flag that restricts the imported database to the first N subject_ids. It works for small tables but is silently ignored for any table whose CSV exceeds THRESHOLD_SIZE (50 MB) — i.e. the big ones (chartevents, labevents, outputevents, emar, etc.). The resulting SQLite database ends up containing every row of those tables instead of the requested subset.

Root cause

The import loop has two paths:

```python
if os.path.getsize(f) < THRESHOLD_SIZE:
df = pd.read_csv(f, dtype=mimic_dtypes)
df = process_dataframe(df, subjects=subjects) # <- filters
...
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) # <- NO subjects=
...
```

process_dataframe only filters when subjects is passed:

```python
def process_dataframe(df, subjects=None):
...
if subjects is not None and 'subject_id' in df:
df = df.loc[df['subject_id'].isin(subjects)]
return df
```

The chunked call omits subjects=subjects, so every chunk of a large table is inserted unfiltered.

Fix

Pass subjects=subjects in the chunked call, matching the small-file path:

```python
chunk = process_dataframe(chunk, subjects=subjects)
```

One-line change. With this fix, --limit N is honored for every table regardless of file size.