New protocol (at the top of the stack) which unbatches OOB message batches, delivering each message in a separate thread. Only active when virtual threads are enabled.

This reduces the costs of processing message batches sequentially, but still harnesses the power of batching (up to the top of the stack).

Therefore probably better than JGRP-2800.

Investigate whether this functionality could be built directly into JChannel.

[1] https://issues.redhat.com/browse/JGRP-2800