Automunge

- inversion now supported for sets with Binary dimensionality reduction
- (as may be applied to consolidation some or all of the boolean integer encoded sets)

- new option for Binary parameter
- (Binary parameter consolidates boolean columns with a macro binary conversion)
- can now pass as a list of source columns and/or returned columns for partial consolidation
- such as to only consolidate a subset of returned boolean columns
- passing a source column header will include all boolean columns returned from that source column, and passing returned column headers allow inclusion of only portion of boolean columns returned from a source column
- note that Binary already has option to either replace or retain the boolean columns inputted to a consolidation
- for this partial consolidation, the default is to replace the target columns
- for option to retain the target columns, can pass the first entry of list as False
- e.g. Binary = [False, 'target_column_1', 'target_column_2']

- a small update
- conformed the normkey retrieval of pwrs and pwr2 postmunge functions to conventions for other returned multicolumn sets with unknown headers
- just trying to make this more uniform, now have single method as standard

- rewrite of pwrs transform for binning by powers of 10
- now closer in composition to pwr2
- also pwrs now accepts boolean 'negvalues' parameter
- defaults to False, when True bins for values <0 are aggregated
- similar rewrite of pwor transform for ordinal binning by powers of 10
- now closer in composition to por2
- pwor also accepts negvalues parameter
- inversion now supported with full recovery for: texd, lb10, lbte, bnrd, lbbn, ord4, ordd, lbor, 101d

- updated UPCS transform to allow retention of nan values for infill with downstream transform
- new transform category 'onht' for one-hot encoding
- similar to 'text', but different convention for returned column headers
- i.e. text returns columns as column + '_entry'
- and onht instead returns columns as column + '_onht_'
- where is integer associated with entry
- new transform category 'Unht' similar to onht with an upstream UPCS
- added inversion support with partial recovery for onht, Unht, Utxt, Utx2, Utx3, Ucct, Uord, Uor2, Uor3, Uor6, U101
- also slight tweak to 'text' transforms to replace an array operation with list for clarity

- found and fixed edge case bug with partial inversion for cases where user includes an excl column with default excl_suffix parameter
- code cleanup throughout, all line breaks capped at single line break
- primarily motivated by reducing number of lines to make code editor rendering a little faster