OS X session termination fixes

[sshimko]

Note this builds on top of pull request #225 and includes the patch from that request.

No idea why, but session initialization and termination is behaving differently on OSX WRT empty bulk transfers. This fix has been tested on OS X Yosemite and Fedora 20.

[sshimko]

@Benjamin-Dobell this pull request needs more testing. I am seeing some unresolvable empty transfer conflicts between the GS5 and Note 3. They could be resolved with a couple device-specific lines in BridgeManager.cpp, but you have managed to avoid that, AFAICT.. Any thoughts on why this is now an issue? LOKE differences on the server side in different bootloaders? USB driver/libusb issues?

[Benjamin-Dobell]

@sshimko Yeah, sorry I really should respond more... I like to wait until I have answers :wink:

I've tested all the current pull requests and they each have merit on some devices but unfortunately break support for other devices. I've gone to fairly great length to try avoid device specific (technically bootloader specific) functionality. Unfortunately detecting different devices isn't actually very straight forward either - all devices share just 3 USB device IDs. The only difference is the config strings they send (but they're quite broken and on several devices they even spell Samsung incorrectly!)