Add a non regression test to deal with buffer types

[hsmett]

Python 2.7:

>>> str(buffer(b"I'm binary"))
"I'm binary"
>>> bytes(buffer(b"I'm binary"))
"I'm binary"

Python 3.6:

>>> str(memoryview(b"I'm binary"))
'<memory at 0x7fd84fc27648>'
>>> bytes(memoryview(b"I'm binary"))
b"I'm binary"

[coveralls]

Coverage increased (+0.4%) to 95.926% when pulling 4a343d6595325e5060e5893a97b5bff613508494 on hsmett:hsmett/binaries into e1d1de0 on romgar:develop.

[coveralls]

Coverage increased (+0.4%) to 95.926% when pulling 14cfab0 on hsmett:hsmett/binaries into d4569cb on romgar:develop.

[romgar]

Why this b64encode?

[hsmett]

Not intended, thanks for noticing.

[hsmett]

Done :)