Datagristle

* gristle_freaker - major updates to enable distributes on all columns to be
automatically gathered through either (all or each) args. 'All' combines
all columns into a single tuple prior to producing distribution. 'Each'
creates a separate distribution for every column within the csv file.
* travisci - added support and started using this testing service.

* Improvement: gristle_slicer - add support for writing data out in the config order
rather than the data order, as well as negative, positive and random stepping.
* BREAKING CHANGE: refactored gristle_validator:
- added examples
- improved error reporting
- replaced obsolete customized jsonschema version with support for the current
version (Draft 7)
* BREAKING CHANGE: Removed gristle_processor.
* BREAKING CHANGE: envvar and config file boolean args with values other than True, true,
t, or 1 will be rejected. This is because they can be very ambiguous and confusing,
in particular with with a pair of args like has-header and has-no-header.
* Improvement: crash reporting is now more consistent, complete and informative
* Improvement: is now enforcing the minimum version of 3.8. It was previously almost
completely working on 3.7 but would occasionally crash on a 3.8 feature.
* Improvement: replaced pyyaml with ruamel.yaml - to better support config generation.
* Bug Fix: fixed pathing on two bash aliases: gristle_determinator & gristle_file_converter
* Bug Fix: fixed bug in handling of obsolete options

* Improvement: the field-names from headers can now be used instead of column offsets
for gristle_sorter, gristle_freaker, gristle_profiler, and gristle_slicer.
* Improvement: The use of the header now follows four simple rules:
- It can be referred to as row 0 when it makes sense - like with gristle_slicer
& gristle_viewer.
- It will be passed through when it makes sense - like with gristle_sorter.
- It will be used to translate field names to offsets for configuration.
- But will otherwise be ignored.
* Bug Fix: gristle_freaker was failing with 0-length files when using col-type=each
* Bug Fix: gristle_sorter was failing with some multi-directional sorts

* Improvement: added gristle_sorter as a script to install in the system so that it is
available to users.

* Improvement: Now supports python versions 3.8 and 3.9.
* Improvement: All csv programs now support envvars and config files for input
and can generate config files.
* Improvement: Programs always autodetect file csv dialect before applying
user overrides - except for piped-in data.
* BREAKING CHANGE: dropped support for python version 3.7
* BREAKING CHANGES to all csv programs:
- Various changes to names of options for consistency, with older versions
caught with an error msg to provides new name.
- Various changes to csv dialect handling for consistency and correct
handling of escapechar, doublequoting, skipinitialspace.

* Improvement: now supports python versions 3.7 and 3.8
* BREAKING CHANGE: dropped support for python version 3.6
* Bumped versions on dependent modules to eliminate vulnerabilities
* gristle_differ
- BREAKING CHANGE: col_names renamed to col-names for consistency
- Fixes --already-unix option bug with file parsing
- Fixes --stats bug with empty files
- Improvement: added ability to use column names from file headers
- Improvement: if a key-col is in the ignore-cols - it will simply be ignored,
and the program will continue processing.
- Improvement: if a key-col is in the compare-cols - it will simply be ignored,
and the program will continue processing.
- Improvement: if neither compare or ignore cols are provided it will use all cols
as compare-cols and continue processing.
- Improvement: CLI help is updated to provide more details and accurate examples of these options.