It probably will never happen. Postgres follows an ideology of explicit and safe...it never tries to assume that you meant to do one thing if there is a possibility that you meant another.

With clusters, there are far too many variables for it to be plug and play. For example, do you want replication, or distributed partitions? If replication, do you want master-slave, or master-master? If distributed partitions, what are you going to trade off: consistency or availability? (That being said, Postgres will likely never change from its CA stance, you'll have to use an alternative distribution like PostgresXL).

Furthermore, clustering isn't likely to be much more than a pareto-tail problem for the next decade or so. It is useful for millions of use cases, and something that only Google's and Twitter's can benefit from is not one of them. If you have the data to justify extensive clustering requirements, I would hope you also have the resources to contribute your patches to Postgres. So far, they all seem to be content solving their own niche problems with niche solutions that slowly bleed down to us mere mortals (Ex. Cassandra), or just buying a solution like EnterpriseDB or Oracle.