Negative CBORTag values are encoding without error in the C encoder

[Sekenre]

Example:

>>> # Pure python module
>>> from cbor2.encoder import dumps
>>> from cbor2.types import CBORTag
>>> dumps(CBORTag(-1, 'val'))
Traceback (most recent call last):
  ...
struct.error: ubyte format requires 0 <= number <= 255

>>> # c module
>>> import cbor2
>>> cbor2.dumps(cbor2.CBORTag(-1, 'val'))
b'\xdb\xff\xff\xff\xff\xff\xff\xff\xffcval'

CBORTag should raise a value error with a negative int