Proposal to update BIP-32

On April 21, 2016, Eric Lombrozo raised a concern on bitcoin-dev regarding the handling of cases where the BIP-32 derivation path is invalid. This issue is compounded by the fact that there is limited software that performs these checks. Furthermore, even if a check is performed, handling the exception can be difficult since skipping may not always be an option. The motivation behind addressing this issue is to enable BIP-32 to be used for non-secp256k1 curves such as the NIST P-256 curve with a chance of 2^-32. An example of an invalid BIP-32 path was found by Jochen at m/28578'/33941 derived from a hexadecimal seed.