Remove unused, untested processors

[maxnoe]

• [ ] fact.features.source.SourceLineTest

  • after an idea by W. Rhode) ;-)
  • implemented by J. Freiwald in 2013
  • code not readable, not commented, not tested))

• [ ] fact.extraction.TimeGradientExtraction

  • implemented by F. Temme, not used, not tested

• [ ] fact.cleaning.MotionCleaning implemented by Lena in 2015, not used, not tested.

• [ ] fact.cleaning.ProbabilityCleaning implemented by @ftemme in 2014, not used, not tested, not working (no @Parameter for the input and output keys

• [ ] fact.features.FillAverageTimeWidth, missing @Parameter for the keys, no docs, not tested, not used.

[jebuss]

I would vote for keeping those but adding unit tests and maybe find some to evaluate those.

fact.features.source.SourceLineTest is a cleaning method suggested by wolfgang. I think it was a good idea but it was never investigate further due to other priorities at that time. Could be an interesting BA.

fact.extraction.TimeGradientExtraction was an Idea for time extraction improvemment based on Sabrinas MA. Same as above.

fact.extraction.TimeOverThresholdTL is a nice tool if you want to investigate what is happening on you time series above a given threshold. However, I agree it would make sense to have an example for this. I think it would not hurt to keep it. Nevertheless, I also don't use it anymore.

[maxnoe]

@jebuss The reasons you state would be reasons for me to go forward removing those processors.

The code of SourceLineTest is not readable and the actual idea is not mentioned anywhere. It also looks like it doesn't work any more.

Same for the extraction things.

I don't think we should keep around ideas that were never tried or finished.

[maxnoe]

Oh, the TimeOverThresholdTL Processor actually head an example, so I keep it. But tests and better docs would be nice to have.

[maxnoe]

I added three more Processors (all cleaning)

[maxnoe]

@jebuss please comment