irt to the read-replica cluster split, wouldn't the read replicat spicedb service

spicedb-replica

in that example still suffer from the same problem? you hit one read replica it says revision 42, then another says revision 43 and boom?