- BREAKING CHANGE: Due to the new imputation of nan-valued features,
rating inference might change slightly. For the "zef18" dataset,
inference was off by about 0.1 on average.
- fix: bad data caused TypeError in numpy.polynomial (25)
- enh: impute nan-valued feature data if corresponding response was 0
- enh: allow empty-valued groups in rating HDF5 file
- enh: when encountering inf values in a training set, replace them
with twice the maxium of that feature