Enumerating tree entries is inefficient due to the root tree OID
Tree entries as returned by various different Gitaly RPCs also return the root tree OID. This is the object ID of the treeish that is at the root of the revision passed to e.g. GetTreeEntry or GetTreeEntries. The problem with this is that we cannot determine the root tree OID without spawning a second process to resolve the object ID. This is inefficient as we not only need to spawn two processes, but every resolved tree entry also requires two roundtrips between each of the processes.
Furthermore, the root tree entries are not actually used by Rails at all. So this effort is ultimately quite pointless and a waste of compute.