[BIP] Normalized transaction IDs

Maxwell suggested using a new P2SH-like scriptPubkey type where the scriptsig is required to be empty. Signatures would not be covered under txid but carried alongside and committed to in blocks in a separate hashtree. Decker noted that this approach was similar to what was described in the BIP, but with the scriptSig used to transport signatures internally to the transactions and stripped when it comes to signing/checking. The only disadvantage to this approach is that it disrupts normal relay handling processes, requiring a different hash or separate mechanism for witness transport.