When provided, computes the stats as if the employee were moved to this target time off policy effective the day AFTER reference_date, so pass the day before the change date - for a change effective on 2026-08-03, send reference_date=2026-08-02. This truncates the source policy's current cycle at that boundary so available_days reflects the prorated balance that would actually transfer or be discarded on the change, rather than the full entitlement. Purely a read - nothing is persisted. Requires the policy timeline feature; without it the result is identical to an un-simulated read.
