Before the match, you have what's called a "constructor" period in the SimpleRobot.cpp template which MIGHT be able to be used to run code in disabled, but my sources haven't tried it yet. If you use the BuiltinDefaultCode.cpp template (iterative template), there's a class for disabled mode. Basically, constructor is the equivalent of Initialize() or IO_Initialization() of the PIC version of WPILib, except I think it runs more than once. Using the disabled mode period, you can run code whenever the robot is disabled, the controller is just introverted; you can read inputs and run code, but you cannot run motors or set outputs until you're enabled.
That being said, that period is the perfect time to figure out where you are and what you're doing with the camera right after the cRIO boots up. With the complexity of the control system this year, there is no reason for your processor to have downtime. You can always be reading a sensor or analyzing visual data to find out where your opponents are and where your alliance members are. That way you know where those robots are already, and can track them when they dart off at the beginning of autonomous and plot a projected course for their trailers to meet very nicely with your moon rocks flying through the air
But seriously, do use the camera.