Definitely agree with Branson's top 10 of challenges with PostgreSQL, especially #3: "Inefficient Replication That Spreads Corruption". Postgres WAL entries are always the bottleneck of a replication stream.

Interesting, that's the one I don't really agree with. Bottleneck, yes, because it is a monolithic stream for the whole database. But corruption?

definitely seen issues in #1 with 32-bit transaction ID and also the problems with connections in PG, but we can sort of manage around these.

Does Yugabyte actually contribute to Postgres or just use their code?

Yes, everything Yugabyte does is open source

Open sourcing the changes doesn't mean contributing back to PostgreSQL: Someone has to put in effort to push changes into upstream PostgreSQL, and if it's not Yugabyte, then IMO they're not contributing back.