The disruption rules allow us to define "disruption windows" where selected shift re-assignments are penalized.

Typically, the disruptions in the near future are penalized more than the later disruptions.

It would be useful to have a finer granularity for the disruption for use cases where the disruption window is insufficient.

One example would be allowing the disruption only when the new shift overlaps with the original (disrupted) shift, or is within a certain scope (e.g. +- 2 hours).

The motivation is that the employees have possibly already arranged their plans based on the original shift, so we want the "new" shift to be as close to the original one as possible.

This could be a hard constraint, for some employees it's a preference, for others it's a requirement.