I agree with the comments which were made on the following points:
- please use consistent naming (`as_bytes` / `from_bytes`, or `asbytes`
/ `frombytes`; my preference goes to the former, especially now that we
have `bit_length`)
- default byteorder should be native, certainly not big endian which is
a small minority amongst today's computers
- you should synchronize with the python-ideas discussion, so that the
final API gets validated more publicly; it would not be very pleasant
for a patch to be committed if discussion were still in flux, and
perhaps with different conclusions as to the API which should be adopted

Besides, your patch has indentation problems (mixed spaces and tabs).