remove .decode('utf-8') for python3 support

[yusufsyaifudin]

this PR intended to fix this issue #14

As mentioned by @Rogdham, I think that we should removed .decode('utf-8') because "Begin with Python 3, all string is unicode object.", based recommendation from answer for this question and answer for this question, it should be removed.

note:
I've tested running the test py.test --cov-report= --cov=ssh-audit -v test and python3 -m py.test --cov-report= --cov=ssh-audit -v test for python3 and have no error.

But, double check should be done because I don't know what the reason behind of using .decode('utf-8'), maybe it to avoid strange behavior for special case (?).

[coveralls]

Coverage remained the same at 56.726% when pulling 8507064 on yusufsyaifudin:master into 76509a1 on arthepsy:master.

[arthepsy]

Thanks for the work and references. You are right, that this bug got introduced to avoid strange case. But as this part of the code wasn't tested automatically (jet), it opened gates for other bug. For more information, read my comment in #14.

I don't think that it's the correct fix to remove .decode('utf-8'), because the source data is bytes and it should be decoded to get back unicode (Python 2) or str (Python 3).

Therefore, I fixed it a bit differently. Also, I added static type checks via (http://mypy-lang.org/) to be sure of the fix. In any case, can You verify that 6b76e68 fixes the bug?

[arthepsy]

This and other encoding/decoding issues are fixed in 1.7.0. Also, multiple of tests added.