Sirad

- Switched from scripts to entry points to support Windows
- Made processing possible when not enough PII exists (e.g., DOB/Names unavailable)
- Fixed leading zeros in zip codes when census coding
- Added dataset name to error message output
- Increased max CSV line size to 1000MB
- Fixed error for null bytes in data

* Added `sirad validate` command (4).
* Consistently use logging instead of print, and set the logging module name with logging.getLogger(name) (6).
* Added threshold on date parsing debug statements.
* Additional address prefixes for censuscoding (mailing, employer1, employer2, employer3).
* Added `--verbose` option.
* Remove quotes from CSV header values in `sirad validate`.
* Use `low_memory=False` in Pandas `read_csv()` calls in `research.py` to address Pandas warnings.
* Additional mappings for non-ASCII characters in `readers.py`.

* Added geocoding component to `sirad research`, which uses address PII to attach Census blockgroup to research files (2).
* Fixed race condition in creating output directories when running `sirad process` with multiple threads (3).

* Improved performance of `sirad research` command.
* Fixed bugs in `sirad research` command for edge cases, typically only found in very small datasets.