in scanner code, replace flock() with select_for_update

[mpdehaan]

the scanner is already horizontally scalable, but you should be able to run more than one scanner per machine.

To fix this, instead of doing a flock(), we should have the scanner do a select_for_update when scanning each individual repo.

This way, multiple scanners can work off the same organization, whether on the same box or distributed, whereas, right now, the system tends to want one scanner per VM.