Reservation Metrics

During this process it calculates the following metrics:

Cancellation Window: Days between check-in and cancellation date (if cancellation date exists).
Booking Window: Days between booking and check-in (for future or completed reservations).
Stay Nights: Number of nights between check-in and check-out (for future or completed reservations).

Inputs

Expects res_df on the workflow context. Required columns: booking_date, check_in_date, check_out_date, and status. Optional: cancellation_date.

Writes back an updated res_df with additional columns: cancellation_window (when cancellation date exists), booking_window (days between booking and check-in for future or completed reservations), and stay_nights (difference between check_out_date and check_in_date for future or completed reservations).

Cancellation window is only calculated when cancellation_date exists on the row. The implementation computes an absolute datediff between check_in_date and cancellation_date.
Booking window and stay nights are calculated when the check_in_date is in the future or the reservation status is one of checked_in or checked_out. Otherwise those fields are set to null.
After computing metrics the runnable checkpoints the reservation DataFrame eagerly and writes it back to the workflow context.

Missing or null dates will yield nulls in the computed columns. The code uses conditional when guards to avoid incorrect values for past or ineligible reservations.
Ensure status values used by the code (checked_in, checked_out) align with your source system semantics; otherwise booking window and stay night computations may be omitted.