(includes a new error code SCRIPT_ERR_SIG_BADLENGTH)
motivation: When Schnorr signatures (64 bytes + 1 hashtype byte) are added, we have the option of leaving CHECKMULTISIG untouched which would mean it would still accept 64+1 byte signatures and interpret them as ECDSA while at the same time, CHECKSIG would accept 64+1 byte signatures and interpret them as Schnorr. This is a sort of awkward ambiguity however, and hopefully it will create less ambiguity, and less risk of consensus break with other implementations, if post-fork *all* 64+1 byte signatures are now Schnorr signatures.