Ticdat

* better `clone` functionality (easier to customize a clone to a slightly different schema)
* fixed deployment problem for sparse Pythons
* minor bug fix in examples
* minor bug fix re: `PanDatFactory` and long Excel sheet names
* Progress tracking for pd pg uploading.

Not putting this release on PyPi. Making it available to anyone who will benefit.

- Fixed a minor bug re: setting up PG schemas.
- Fixed a minor bug re: initialing `PanDat` objects with empty lists for the named arguments.
- Made the csv example data a little more challenging to import.

Not putting this release on PyPi. Making it available to anyone who will benefit.

- Performance improvement for Excel reading with `PanDatFactory`
- Uses `openpyxl` for reading .xlsx files (latest `xlrd` no longer works for this).
- This includes some work to prune trailing all-empty rows, which were previously auto-pruned by the `xlrd` code.
- Ancient branches with example code finally got merged in.

Not putting this release on PyPi. Making it available to anyone who will benefit.
* Fixing a slow data load problem for `TicDatFactory` and Excel files. Should have been fixed with 0.2.19.6 but I overlooked Excel.

Following summarizes all changes since the 0.2.19 Release.
* OpenX action support
* Row predicates can be empowered to look at other rows from other tables.
* Row predicates can create individualized messaging.
* `max_failures` allows integrity checking to fast fail on dirty data.
* Functionality for restricting the solve and the actions to subschemas.
* Postgres date support.
* Run time performance improvement re: automated data munging.
* `find_data_row_failures` will treat row predicate exceptions as dirty data.
* `standard_main` has `case_space_table_names` argument.
* Addresses the irritating "is with a literal" warning.
* Better documentation.
* Better support for non-pandas installations.
* `replace_data_type_failures` for `PanDatFactory`.
* better handling of postgres reserved words.
* unneeded columns purged from postgres pandas write back.
* Better error messages if solve/action returned objects not bool-able (for ex. `DataFrame` returned by accident).
* minor bug fix in `ticdat.Slicer`, `good_tic_dat_object`, `add_cplex_listener`.

Not putting this release on PyPi. Making it available to Enframers and anyone else who will benefit.

- Better error messages if solve/action returned objects not bool-able (for ex. `DataFrame` returned by accident)
- better performance by fixing some dumb time wasting stuff on data ingestion and pre-writing
- `standard_main` has `case_space_table_names` argument
- `max_failures` with integrity check failures - allow integrity checker to fast-fail on dirty data.
- `ticdat.Slicer` very corner case bug resolved
- minor flaw in `good_tic_dat_object` fixed
- PostGres date support