I think the biggest decision is how much do you bounce off the wall and lose 7-10% or more as well as the loss of intensity on other side vs. improving overall spread of light.
finding that negligible returns of one direciton and how much is lost on other side at that point? a simple light meter can give relative intensity.
if the time to reverse directions causes uneven dirstribution of time, that's something to counteract with programming, if it is that sophisticated/robust of an option. If not, 2 well-timed, fixed locations that don't causes significant loss on either side might average out best. a light meter can help. par meter and some resulting maps would help more.