So length defaults to i.bit_length():

    >>> s = Bits(uint=44)
    >>> s.len
    6

Can also be done for uintle, uintbe, uintne (rounding up to whole byte).

Not sure it makes sense for int etc., although bit_length() does work for 
negative integers, so it might be fine.

Note that (0).bit_length() == 0, which is something to think about - should 
zero need at least one bit or should we let it be an empty bitstring?