* penalty of applying the batches is spread across threads so that the amortized cost is slightly
   * higher than performing just the operation without enforcing the capacity constraint.
   *
   * This implementation uses a per-segment queue to record a memento of the additions, removals,
   * and accesses that were performed on the map. The queue is drained on writes and when it exceeds

   * penalty of applying the batches is spread across threads so that the amortized cost is slightly
   * higher than performing just the operation without enforcing the capacity constraint.
   *
   * This implementation uses a per-segment queue to record a memento of the additions, removals,
   * and accesses that were performed on the map. The queue is drained on writes and when it exceeds

    public static final String CONSTRAINTS_AssertFalse_MESSAGE = "{constraints.AssertFalse.message}";

    /** The key of the message: {item} must be true. */
    public static final String CONSTRAINTS_AssertTrue_MESSAGE = "{constraints.AssertTrue.message}";

    /** The key of the message: {item} must be less than {value}. */
    public static final String CONSTRAINTS_DecimalMax_MESSAGE = "{constraints.DecimalMax.message}";

- Added SchedulingConstraints to express TAS constraints for pod group scheduling behind TopologyAwareWorkloadScheduling feature gate.