Extend "average accelerometer" measure to provide more features

[bardram]

To further address #336, we want to extend the average accelerometer measure (PR #342, #341) to contain more features.

Using these articles as inspiration:

• Feature Engineering on Time-Series Data for Human Activity Recognition

[bardram]

The statistical features:

13. number of peaks
14. skewness
15. kurtosis

could not be implemented (they assume some sort of probabilistic distribution).

Peak detection could be implemented, by using this Robust peak detection algorithm (using z-scores).

[bardram]

Published as 1.3.4.

Accelerometer features are now like this:

{
 "sensorStartTime": 1697800468891780,
 "sensorEndTime": 1697800478903568,
 "data": {
  "__type": "dk.cachet.carp.accelerationfeatures",
  "count": 55,
  "xMean": -0.002724063667384061,
  "yMean": -0.002637693015011874,
  "zMean": -0.16934890747070314,
  "xStd": 0.013855562205479371,
  "yStd": 0.01849891120176733,
  "zStd": 0.011855636444812601,
  "xAad": 0.011472705031229444,
  "yAad": 0.013146298109007278,
  "zAad": 0.008665431629527699,
  "xMin": -0.03441151976585388,
  "yMin": -0.06160842627286911,
  "zMin": -0.20050811767578125,
  "xMax": 0.022999465465545654,
  "yMax": 0.04248298704624176,
  "zMax": -0.12859725952148438,
  "xMaxMinDiff": 0.057410985231399536,
  "yMaxMinDiff": 0.10409141331911087,
  "zMaxMinDiff": 0.07191085815429688,
  "xMedian": -0.0013580620288848877,
  "yMedian": -0.0037148892879486084,
  "zMedian": -0.16947650909423828,
  "xMad": 0.010873883962631226,
  "yMad": 0.010325059294700623,
  "zMad": 0.004851341247558594,
  "xIqr": 0.019643381237983704,
  "yIqr": 0.020682185888290405,
  ...
}