Scikit-digital-health

Changes
---------

- Multiple bug fixes
- Multiple algorithm enhancements

Rename to scikit-digital-health (skdh)

Minor bug and typo fixes
Updated testing framework

Changes

- Activity classification has been fleshed out and now generates endpoints for sedentary/light/moderate/vigorous levels, as well as excluding sleep from the analysis (but reporting results per epoch during sleep if sleep is available)
- sleep now has its own internal sleep-only wear detection
- GeneActiv bin reader can now handle differing sampling frequency values between the header and data blocks
- Axivity reader now can handle multiple bases and periods
- fs is now a valid input, but if not provided is still computed from the timstamps

Added additional date/time information to sit-to-stand output
Added start of activity metrics

Changes:

- Added wear detection algorithm for accelerometer only

Changes:

- Add CalibrateAccelerometer method to auto-calibrate already collected at-home data
- Add rolling statistics methods in utility submodule
- Fix gait metric computation when using autocovariance functions (Gait Symmetry Index, Stride regularity, etc)
- Use median for summaries of previously computed metrics (GSI, regularity, etc)
- Add orientation correction (Moe-Nilsson 1998) as optional to Gait processing
- ReadBin now returns a temperature array that is the same length as the time and acceleration arrays
- Standardize file reader function/class names
- Rolling mean and standard deviation now use almost no additional memory during computation
- Fixed a bug in the rolling moment extensions that was causing segfaults (return dims freed before final use)