Stale versions of nodes do not need to be written. They can instead be substituted by checking if the next version has been written.

We should (1) stop writing stale nodes, and (2) we should also stop checking for the existence of them in lookup and key history proofs.

For (1), this will not be compatible with existing deployments, so we can stop writing stale nodes in the experimental configuration but keep writing them in the whatsapp_v1 configuration.

For (2), we can turn off lookup and key history proof checks for stale nodes since these are not persisted.