[BigQuery, BigQuery Storage]: Add option to use BigQuery Storage API to download results in BigQuery DB-API

[tswast]

Is your feature request related to a problem? Please describe.

In projects like superset, that use the SQL Alchemy connector, it can be quite slow to download large query results. The BigQuery Storage API speeds this up for to_dataframe / pandas, but not when used via the DB-API / SQL Alchemy. See: googleapis/python-bigquery-sqlalchemy#41

Describe the solution you'd like

When creating a DB-API Connection, provide a way to supply a BQ Storage client, in addition to a BQ client. Use this client to download results for the relevant methods in the Cursor object.

Describe alternatives you've considered

Could have a use_bqstorage_api option, but this would be inconsistent with the current constructor, which expects a client.

/cc @yiga2

[tswast]

The proposal here is to add a bqstorage_client argument to the Connection class:

https://github.com/googleapis/google-cloud-python/blob/b387134827dbc3be0e1b431201e0875798002fda/bigquery/google/cloud/bigquery/dbapi/connection.py#L21-L26

Instead of always calling list_rows to fetch the query results

https://github.com/googleapis/google-cloud-python/blob/b387134827dbc3be0e1b431201e0875798002fda/bigquery/google/cloud/bigquery/dbapi/cursor.py#L215

attempt to call the BQ Storage API first. This can be modelled after to_dataframe, where an iterable over rows is abstracted away.

https://github.com/googleapis/google-cloud-python/blob/b387134827dbc3be0e1b431201e0875798002fda/bigquery/google/cloud/bigquery/table.py#L1407-L1436

Just as with to_dataframe, there will be cases where the BQ Storage API will fail, but tabledata.list will succeed, so we'll need to fallback to tabledata.list even when a BQ Storage API client is available.

Changes from to_dataframe logic:

• Request only a single stream, as the DB-API isn't well suited for multi-threaded reads.
• Iterate over rows rather than pages. https://github.com/googleapis/google-cloud-python/blob/b387134827dbc3be0e1b431201e0875798002fda/bigquery_storage/google/cloud/bigquery_storage_v1beta1/reader.py#L146