Libhxl

* module now has a __version__attribute, per PEP 396
* validation enhancements:
* major refactor of the hxl.validation module for better testing and maintainbility
* new default validation schema
* ability to generate a JSON validation report
* validation schema rules can support multiple tag patterns
* test for spelling inconsistencies
* test for numerical outliers
* filter changes
* refactor hxl.filters.CacheFilter to preserve source row numbers
* hxl.filters.RowFilter no longer ignores empty cells
* when multiple columns match a row query, it will succeed with at
* fix bug in the "is" operator for row queries
* handle more types of Google Sheets URLs
* when times are attached to dates, date parsing will still succeed
* new external module dependency: python-io-wrapper

- hotfix for a rare bug in date parsing

* added wildcard support to tag patterns, so that we can use patterns like "*" or "*+f-children"
* removed obsolete Python2 compatibility code
* added source_row_number and source_column_number to support validation
* revamped date handling to support partial dates like "2018-01" or "2018", and also special notation like "2018W05" or "2018Q1"
* row queries support is (not) min and is (not) max, including for numbers and dates
* added min() and max() methods to hxl.model.Dataset

Major enhancements to the validation engine:
* now accepts all parseable date formats
* new valid_unique constraint (single value or compound key)
* new valid_correlation constraint (e.g. make sure that adm1+name is always consistent for any given value adm1+code)
* new valid_datatype+consistent test to to infer datatypes in a column without explicit rules
* new valid_value+whitespace test
* suggests spelling corrections when validating against a list with valid_value+enum or valid_value+url
* new valid_value+whitespace test detects irregular whitespace

For more details, see the CHANGELOG

**Note: Python2 no longer supported.**

Core enhancements:
- Added Python logging support (will expand in next release).
- can now open data from a CKAN dataset URL (will look for the first resource)

Filter enhancements:
- clean filter can normalise lat/lon
- column filter can remove all columns without HXL hashtags
- 'patterns' parameter is now optional for JSON-encoded count filter

Bug fixes:
- Google Sheets now open properly from a CKAN resource URL
- Restore support for preserving original attribute order.

Interim bug-fix release. A badly-formatted date was sending an exception to the top level and halting processing.

Enhancements to filters:
- merge-data looks for key matches in all candidate columns
- cut-columns can remove all columns without HXL hashtags
- clean-data allows number formatting strings like ``0.2f``

Try to recognise JSON data even when it doesn't have a JSON MIME type or ``.json`` extension.

Generate JSON lists of objects as described in the [HXL 1.1 beta spec](http://hxlstandard.org/standard/1_1beta/tagging/#tagging.json.objects)

See the CHANGELOG for more details.