Unpin pandas-gbq and remove unused code (apache#21915)

* Unpin ``pandas-gbq`` and remove unused code

`BigQueryPandasConnector` was previously used by `BigqueryHook.get_pandas_df`. This was fixed in apache@ad308ea (**in 2018**). However we forgot to remove `BigQueryPandasConnector` which is age-old code (2016) and used private methods.

2016 code - https://github.com/apache/airflow/pull/1452/files

airflow/contrib/hooks/bigquery_hook.py

@@ -24,7 +24,6 @@
     BigQueryConnection,
     BigQueryCursor,
     BigQueryHook,
-    BigQueryPandasConnector,
     GbqConnector,
 )
 

airflow/providers/google/cloud/hooks/bigquery.py

@@ -46,11 +46,7 @@
 from googleapiclient.discovery import Resource, build
 from pandas import DataFrame
 from pandas_gbq import read_gbq
-from pandas_gbq.gbq import (
-    GbqConnector,
-    _check_google_client_version as gbq_check_google_client_version,
-    _test_google_api_imports as gbq_test_google_api_imports,
-)
+from pandas_gbq.gbq import GbqConnector  # noqa
 from sqlalchemy import create_engine
 
 from airflow.exceptions import AirflowException
@@ -2193,28 +2189,6 @@ def run_query(
         return job.job_id
 
 
-class BigQueryPandasConnector(GbqConnector):
-    """
-    This connector behaves identically to GbqConnector (from Pandas), except
-    that it allows the service to be injected, and disables a call to
-    self.get_credentials(). This allows Airflow to use BigQuery with Pandas
-    without forcing a three legged OAuth connection. Instead, we can inject
-    service account credentials into the binding.
-    """
-
-    def __init__(
-        self, project_id: str, service: str, reauth: bool = False, verbose: bool = False, dialect="legacy"
-    ) -> None:
-        super().__init__(project_id)
-        gbq_check_google_client_version()
-        gbq_test_google_api_imports()
-        self.project_id = project_id
-        self.reauth = reauth
-        self.service = service
-        self.verbose = verbose
-        self.dialect = dialect
-
-
 class BigQueryConnection:
     """
     BigQuery does not have a notion of a persistent connection. Thus, these

setup.py

@@ -355,9 +355,7 @@ def write_version(filename: str = os.path.join(*[my_dir, "airflow", "git_version
     'grpcio-gcp>=0.2.2',
     'httpx',
     'json-merge-patch>=0.2',
-    # pandas-gbq 0.15.0 release broke google provider's bigquery import
-    # _check_google_client_version (airflow/providers/google/cloud/hooks/bigquery.py:49)
-    'pandas-gbq<0.15.0',
+    'pandas-gbq',
     pandas_requirement,
     'sqlalchemy-bigquery>=1.2.1',
 ]