Automunge

new transformation categories to convert continuous numerical sets into bounded range:
dxdt, d2dt, d3dt, dxd2, d2t2, d3d2, nmdx, nmd2, nmd3, mmdx, mmd2, mmd3
where e.g. dxdt is rate of change (velocity), d2dt (acceleration), d3dt (jerk)
assumes unshuffled data with consistent time steps
dxd2 comparable but evaluates delta of average of two rows with average of two preceding rows
such as to denoise the data
nmdx is comparable but a z-score normalization is performed prior
mmd is comparable but a min-max scaling is performed prior

update version number

revision to new import procedure

trying to get imports set up, forgot to include init duh

restructuring imports to pip install Automunge etc. I think this will work fingers crossed.

- new method for retrieving normalization parameters in text category, now one hot encoding can be performed downstream of prior transformations
- prior method reframed as support function for other functions' use