Option to specify or auto-detect desired order of sleep 'guider' methods to be used

[vincentvanhees]

This is something I currently do not have time/funding to work on, but it would be good to have in the future.

At the moment we specify one sleep detection method and if that fails there is some hard-coded backup mechanism that is a bit tailored to wrist-worn sensors. For example: If sleeplog is used but missing => use HDCZA => If could also not be derived then use L5+/- 12 hours.

However, this is not logical. For example, if we specify guider "NotWorn" then the fall back option would be L5+/- 12 hours, while maybe we want this to be HDCZA for studies with wrist placement. Or another example, if we specify HDCZA and it is clear from the data that sensor was not worn during the night then it may be best to fall back on "NotWorn".

To address this I think we need:

• Clear map of all possible scenarios and where classification may go wrong.
• Make part 3 more clever in deriving backup guiders when the default is likely to fail.
• In part 4 select the most appropriate guider.

[vincentvanhees]

Commit link above shows my fix for this issue in relation to NotWorn guider.

Or another example, if we specify HDCZA and it is clear from the data that sensor was not worn during the night then it may be best to fall back on "NotWorn".

This is actually not needed, because HDCZA can already consider nonwear to be part of SPT.

[vincentvanhees]

re-open because I am currently making further updates to this.