Add option to let all sib count towards sleep when overlapping with TimeInBed

[vincentvanhees]

At the moment TimeInBed window type means that only sib are counted as sleep if they entirely overlap with TimeInBed motivated by the idea that sleep that overlaps with the edge of being in bed (self-reported) cannot be sleep.

In a project I am working on at the moment we see that actually it is more likely that the sleep diary is incorrect and that we should count those sibs as sleep.

To not break consistency with the past I will add a new option to change this.

This means that sleep latency can be negative.
Plan is to report negative latency at night level but ignore at person level. This does not affect any of the other sleep estimates only the latency estimate.

Next:

• Add option to control when sibs are counted towards sleep
• Deal with potential for negative latency, see above.
• Add extra statistics to monitor how many nights had negative latency and were excluded from the person level latency estimate
• Test in real dataset
• Add control over whether this only applies for beginning or end of the night
• Add extra statistics on sib during the hour before and after SPT as an indicator for whether sleep may have been missed. => better to do this in part 5 if we do it, but for now just visual inspection is best as way of method evaluation

[vincentvanhees]

Status: This is now part of isseu967_selfreport to be test in that project, and I will merge all the work as one PR in the new year.