For testing purposes.

For theoretical future operational purposes I'd imagine they'd still want to be able to use this when planes are flying. But could likely track planes and adjust the beam slightly to avoid them, or just disable the system for the rare couple of seconds when a plane is in the exact wrong place (in front of the beam).

It makes sense to do that for a testing site and wait for lightning to happen. How does it work for general purpose?

They can’t just say “in three microseconds we’re restricting airspace —get out now! Ooops, too late!”