Unless I'm missing something, I believe they can do everything they need with either ramp or mainline scanning or a combination of the two. As long as they have enough of them that they can always tell from which ones you passed through where you got on and where you got off, the correct toll can be computed and charged.
Moreover, number of
readers doesn't really depend on pattern.
Number of
reads, on the other hand, does - and it may easily turn out IT savings are not insignificant.
If it was up to me, I would link the new pattern to the expectations of exit reconfiguration. For example, exit 24 is an eyesore, and I would expect reconfiguration will happen at some point. Gantries on remote approaches make sense. Exit 29, I believe, has no expectations for reconfiguration, so using existing infrastructure makes more sense.