urllib.parse.quote_plus will ignore its encoding and errors arguments if
its input string has a space in it.

Intended behaviour:
>>> urllib.parse.quote_plus("\xa2\xd8 \xff", encoding='latin-1')
'%A2%D8+%FF'
Observed behaviour:
>>> urllib.parse.quote_plus("\xa2\xd8 \xff", encoding='latin-1')
'%C2%A2%C3%98+%C3%BF'
(This just uses the default UTF-8 encoding).

Attached patch with test cases. This only affects Python 3.x (the 2.x
branch has no encoding/errors argument).