When calculating compliance with work-hour rules (rolling averages, consecutive days, rest periods), days marked as absence — vacation, sick leave, public holidays — should be excluded from the rolling window entirely. Rather than treating those days as zero-hour work days, the window should "skip over" them and extend further back in time to collect the required number of qualifying days or weeks.

Problem

Currently, unavailability markers cause a shift employee's compliance window to include a period where they weren't working at all. This produces misleading results:

An employee on vacation in week 10 who is now in week 12 gets evaluated over weeks 9–12 under a 4-week rule. Week 10 contributes zero hours, making them appear to have worked far fewer hours than they actually did across their real working weeks.

Consecutive-day rules may fire incorrectly, because an absence day acts as a "break" in the sequence in ways that don't reflect real scheduling intent.

For the 4 week period with week 10 as vacation, this would instead evaluate work and breaks over weeks 8, 9, 11, and 12, as if week 10 did not exist.

The existing "unavailability" concept doesn't solve this because it still participates in the time window. What's needed is a distinct concept — call it an excluded period or absence tag — that causes the window to contract around it.