[WIP] drop julia < 0.5 and prepare for 0.6

[CarloLucibello]

trying to address #23

[coveralls]

Coverage decreased (-3.5%) to 84.371% when pulling 22dbcbc602f593fc21048025539f61bd2cf6ee3b on CarloLucibello:j06 into a0c0da6cbaa98ed707d4290c899589e7833dbec5 on andrewcooke:master.

[coveralls]

Coverage decreased (-4.5%) to 83.38% when pulling e90a9c901386fac9db8e18e1dd604b4dfe84507b on CarloLucibello:j06 into a0c0da6cbaa98ed707d4290c899589e7833dbec5 on andrewcooke:master.

[CarloLucibello]

@andrewcooke @sbromberger can someone continue from here or give me a hand?

On julia 0.6 there are still a lot of tests not passing, and some passing but giving weird results.

Also due to the new Channel sintax this is not working on 0.5 right now and I'm not sure 0.5 compatibility can be kept. This is a minor problem, maybe it is just better to drop 0.5 to have the most maintanable thing we can have

[sbromberger]

Yeah, we're seeing similar issues in LG right now. Our solution is to move 0.6 compatibility into its own branch that will not be backwards-compatible with 0.5.

[CarloLucibello]

@andrewcooke if you find some spare time please continue the update work from here. There are a few issues left that I couldn't manage to solve, shouldn't take much though.

Bye Carlo