rlp.decode_to_bytes doesn't raise with negative decoded len

[ClementWalter]

Using rlp.decode_to_bytes(encoded_bytes=b'Q59:\xba\xf4\xda\x05\xb7') returns b'' because it silently slices the input data with a negative len

def decode_to_bytes(encoded_bytes: Bytes) -> Bytes:
    """
    Decodes a rlp encoded byte stream assuming that the decoded data
    should be of type `bytes`.

    Parameters
    ----------
    encoded_bytes :
        RLP encoded byte stream.

    Returns
    -------
    decoded : `ethereum.base_types.Bytes`
        RLP decoded Bytes data
    """
    if len(encoded_bytes) == 1 and encoded_bytes[0] < 0x80:
        return encoded_bytes
    elif encoded_bytes[0] <= 0xB7:
        len_raw_data = encoded_bytes[0] - 0x80

----> here len_raw_data = -47

        if len_raw_data >= len(encoded_bytes):
            raise RLPDecodingError
        raw_data = encoded_bytes[1 : 1 + len_raw_data]
        if len_raw_data == 1 and raw_data[0] < 0x80:
            raise RLPDecodingError
        return raw_data

Some other examples

• encoded_bytes=b'~\xbc\xc5^\xbe\xff' gives len_raw_data = -2 and do return a bytes

I think that if len_raw_data >= len(encoded_bytes) should also check for negative lengths.