pandas.DataFrame.to_sql using connector doesn't write the last row

There is a strange bug - when I try to use this connector to insert rows of pandas.DataFrame to ClickHouse.

```
import sqlalchemy as sa
import pandas

CLICKHOUSE_DSN = '***SECRET***'
ch_db = sa.create_engine(CLICKHOUSE_DSN)

pdf = pandas.DataFrame.from_records([
{'year': 1994, 'first_name': 'Vova'},
{'year': 1995, 'first_name': 'Anja'},
{'year': 1996, 'first_name': 'Vasja'},
{'year': 1997, 'first_name': 'Petja'},
])

pdf.to_sql('test_humans', ch_db, if_exists='append', index=False)
```

When I run it, I can see first 3 rows of pdf in table, but not the last one :-)
When I use postgres connector, or, this alternative clickhouse connector https://github.com/xzkostyan/clickhouse-sqlalchemy all works fine.

Start by reproducing the example with pandas.DataFrame.to_sql, a SQLAlchemy engine, and the ClickHouse connector, then compare the inserted rows with the four source records. Trace the connector's insert path and verify why the final row is omitted; done means all four rows appear, as they do with the PostgreSQL and alternative ClickHouse connectors.

Written by the indexing model from the issue text.