WriteToBigQuery: Schema auto-detection is not supported.

This is regarding the WriteToBigQuery connector which is supposed to support SCHEMA_AUTODETECTION but throws an error indicating that Schema auto-detection is not supported for streaming inserts into BigQuery (Only for File Loads) even when batching from a file.

SCHEMA_AUTODETECTION should be supported as per instructions (or consider adding more context to the instructions):  [https://beam.apache.org/releases/pydoc/2.15.0/_modules/apache_beam/io/gcp/bigquery.html](https://beam.apache.org/releases/pydoc/2.15.0/_modules/apache_beam/io/gcp/bigquery.html)
 
 I am proofing out a CSV ingest workflow using the Direct Runner and Interactive Runners - the results are the same.  
 

Working in Python 3 (3.7.8 packaged by conda-forge).  

Pip installed via install apache_beam[gcp]

apache_beam.__version__ = 2.23.0

Imported from Jira [BEAM-10666](https://issues.apache.org/jira/browse/BEAM-10666). Original Jira may contain additional context.
Reported by: yaquino.

Start with the Python WriteToBigQuery connector and the linked BigQuery documentation for SCHEMA_AUTODETECTION. Reproduce the error using a file-batched workflow with the Direct or Interactive Runner, then verify that schema auto-detection works as documented or that the instructions clearly describe the limitation.

Written by the indexing model from the issue text.