Bcpandas

Major Changes:
- Better docs
- Suppressed output from command line tools in `__init__.py`
- In `to_sql` with `if_exists='replace'`, now uses `pandas`'s own internal code to create the new SQL table from the dataframe, including preserving the column types and other schema information. Big improvement over the default NVARCHAR columns that was until now.
- `SqlCreds` now creates a `sqlalchemy.Engine`, and can also be created from one.
- `bcp` now outputs STDOUT in real time, instead of waiting until the process is complete
- Bug fix in creating BCP format file if the column name was not a string
- Started working on benchmark tests, using `pytest-benchmark`
- Added dependencies to support the `sqlalchemy.Engine` object

* Added flags to BCP and SqlCmd to `SET QUOTED_IDENTIFIERS` to `ON`, needed for Azure SQL Data Warehouse

* `to_sql` attempts to use several possible delimiters, to fix BCP not escaping delimiters with quotes
* Bug fix in `to_sql` for `if_exists` param if `fail` is passed
* Added Caveats and Limitations section in README
* Insulated exceptions with custom exception class, for easier testing and debugging
* Updated tests to use Docker for local SQL Server instance
* Fixed bugs in tests

Renamed SqlCreds method `from_quantity` to `from_engine`, deployment automation