Update to python 3.7 and latest nuklear

[creikey]

Everything except the overview translated to python seems to work
perfectly. The biggest change between versions was that now with
every string on cffi you have to do "test".encode('utf-8') instead
of just using a string in a cffi function like "test". I have not
found a way to mitigate this issue so far. This change in encoding
practice with cffi is the source of almost all breaking in
overview.py. I used a vim regex command to add encode('utf-8') to
all strings that aren't doc strings, but there are several issues
still present and undiscovered, and can only be discovered by
running overview and pressing on random buttons.

[creikey]

Seems not to be building on travis...

[creikey]

Ok everything seems to be building now, had to remove some stuff with sed in a makefile

[creikey]

So it turns out that you can use b"my string" instead of "my string".encode('utf-8'). This should make development a little easier with the new changes

[nathanrw]

Hi,

Thanks for contributing this! Python 3 support would be fantastic.

I've not got a vast amount of free time at the moment so it might be a little while before i can look at merging your changes.

A few things stick out at me that probably mean i cant merge it in its current form

• upgrading nuklear and python 3 support are 2 different things, so we should do 1 at a time. Python 3 support is less likely to cause downstream breakage if done correctly, whereas a major library upgrade could do who knows what.
• we should endeavour to retain python 2 support as well as support python 3.
• would like to avoid dependency on unix commandline tools (the fact that the preprocessed header is committed to source control is really a bodge and id prefer to avoid relying on it)
• its best to avoid making unnecessary changes e.g. reformatting code without otherwise changing it

I'll try to find the time to take a closer look as soon as i can.

Thanks

Nathan