use unique_lock instead of mutex and re-wrote IterateEvents

[CalebWhiting]

Fixed crash as seen on issue #44

To be completely honest, I'm not a C++ savant. I don't know the root-cause of the crash except that it was this line

alt1-electron/native/os_x11_linux.cc

Line 266 in ae28add

while(!it->done) it->condvar.wait(lock);

And re-writing the IterateEvents method to use futures fixed it for me.

[Fargekritt]

Had the same issue and this fixed it for me aswell.

[Arroquw]

Tested this on my machine, somehow this causes the alt1 window to reset OS initiated moves and resizes of the window and only honours these actions if they are done through the window's own means to move or resize.

It does, however, fix sporadic crashes that I was getting.

[flut2]

The issue here is that condvars is modified both by the record and the window thread without being appropriately protected, as they share the vector.

Potential fixes are:

• Just moving eventMutex.unlock(); to the bottom of the function (slightly overprotective, this isn't the only thing eventMutex guards)
• Not making condvars global, construct it on the spot, as it gets cleared anyways (slightly more allocations though)
• Adding a new mutex that protects condvars, this is likely to be the most performant, but also marginally more involved when performance shouldn't really be a problem here (the events are too infrequent)

Though it doesn't seem like the repo is maintained anymore, just left this here for the various forks.

[Arroquw]

Just a heads up, I incorporated this PR into my fork:

https://github.com/arroquw/alt1-electron