Pyathena

* Change connection options to be retrieved from the dialect class instead of the bind parameter 352
The following error in the Dataframe.to_sql method of Pandas 1.4 has been fixed.

sqlalchemy.exc.CompileError: The location of the table should be specified by the dialect keyword argument `awsathena_location`

* Suppress log output when failure to retrieve metadata from SQLAlchemy 353
Thanks plazmakeks 326
* Update botocore and boto3 dependencies 350
Thanks aaronclong 341

* Change the Parquet library used by Pandas to allow choice of PyArrow or fastparquet 343
The PyArrow dependency is no longer required in the extra package Pandas.
If you use the unload option, you must install the extra package with `PyAthena[Arrow]` or `PyAthena[fastparquet]`.
Thanks aaronclong 341
* Change the default value of na_values in AsyncPandasCursor from None to empty character 347
Thanks dblado 346
* Support for specifying ExternalId in AssumeRole 348
Thanks dblado 344

* Update s3fs dependencies 342
Thanks nikhil27472 340

* Update botocore and boto3 dependencies 337
Thanks zachliu 336

* Fix rendering of LIMIT / OFFSET clauses in SQLAlchemy to work with version 1.3 and lower 335
Thanks mdeshmu 334

* Support credentials with profile names in ArrowCursor and PandasCursor 332
* Add region name to boto3 session 333