Incorrect bits passed to EMSA-PSS-ENCODE

[blake-watkins]

Apologies if this is the incorrect place to raise this. I was looking at the draft and the number of bits passed to EMSA-PSS-ENCODE here seems wrong (section 4.2 step 1): encoded_msg = EMSA-PSS-ENCODE(msg, bit_len(n))

The argument needs to be bit_len(n) - 1 to ensure that the encoded message is less than the modulus. I think the minus one was there originally but got left off as part of the change from bytes to bits in #173.

The python implementation is working despite passing the modulus bit length to EMSA-PSS-ENCODE because the implementation of EMSA-PSS-ENCODE subtracts one from that argument immediately (which isn't specified in the RFC

def EMSA_PSS_ENCODE(kBits: int, msg: bytes, sLen: int, salt: bytes = None) -> bytes:
    m_hash = H.new(msg).digest()
    hLen = H.digest_size

    emBits = kBits - 1 ...