I assume this cannot affect the prioritisation of the Coinbase transaction that constitutes p2pool miner's reward payments? Can you clarify this?
The coinbase tx itself? No (it would be in the block you solve, nothing can change that).
Spending the outputs of the coinbase tx? Yes.
p2pool could implement BIP32 or in the interim given that p2pool nodes also run bitcoind could just implement a direct getnewaddress() RPC call after any successful block to ensure future blocks use a unique address.
I'm not sure which I prefer, an RPC call to generate a new random address, or supplying a keychain of addresses that are predetermined... I guess it depends on whether the keychain appears on the blockchain or is made public through the p2pool network. I can't think of a reason for the latter to happen, I don't see how other p2pool nodes need the keychain itself, they should only need the latest iteration of address from within the keychain sequence. But this all depends on how p2pool and BIP32 neccesarily operate (and how the pair of them operate together).
forrestv, LukeJr, anyone else who can shine a light: we welcome your comments