Since there is a need for a PLC, I would consider bringing in the humidity too and let the PLC run the exhaust fan. Then you can run the fan on a periodic basis whether the humidity is high or not and that can help a great deal getting the CL2 out of the room. Also, you can report the humidity back to them. We have a couple of sites where they have humidity monitoring and they can tell right away when a pump seal is leaking if there is a spike in the trend not due to weather.
If you put the pressure sensors in the PLC and let the PLC run the VFD command, there are more advantages than disadvantages in my opinion. The big disadvantage is that if the PLC goes kaput, then you can't run the VFD in hand with pressure control. The advantages include being able to do whatever your imagination conceives with the pump speed. You could limit the pressure command, use steps to incrementally raise it, the sky is the limit.
I would not be surprised if you could go tweak some settings, validate the sensor wiring and make the junk behave with just a few hours time on site. I have seen situations like this. One place had replaced a 150psi sensor with a 200psi sensor and didn't "tell the drive" so yeah, it was running way too high. They also had a faulty gauge so it was not obvious. Another site was all wired right and the VFD was programmed right, but it was in the wrong mode so running at a fixed speed instead of using the sensor.
I like to use Micrologix PLCs with the 1761-NC01 board and Modbus comms to the drives. Then you can tell if the keypad is in auto or hand mode, monitor motor current, and everything else about the drive you might want. This may not seem important but will pay off when they call you in the middle of the night and you can log into their office or look at a picture of a screen with drive details to tell you what is really going on without taking a road trip. It also can cost less in hardware than analog signals and make wiring very simple.
The PID programming and adjustment for pressure control with the sensor on the discharge is not terribly challenging, so the risk of overshooting is low if you cover all your bases not the least of which is detecting a failure of the sensor. With that said, I have seen every possible mode of failure of pressure sensors in the last year. I have one I keep on the bench that puts out 20.5 milliamps at all times. I have another one that "froze" at 16.5mA. Programming to detect these possibilities is a good idea.
Also, I think I saw check valves in the pictures on the pump discharge lines, so (if they are working) that should prevent backdriving the pumps and backflow into the sump.
One more thing, in rooms with chemicals, especially CL2, I try to use poly or fiberglass enclosures. I have had great service and pricing from Wistex for boxes up to 20 x 16 which is adequate for many of my small PLC/RTU panels.
https://www.wistexllc.com/enclosure...aque-cover-wall-mount-enclosure-assembly.html