Ok. A pwm splitter has 2 cables. 1x cable has 4 wires, a 12v+,G, tach and pwm signal. The other cable has 3 wires, 12v+, G, pwm signal.
When you put 2x pwm fans on a splitter, the header only receives 1 tach reading, there's only 1 tach wire, but sends the pwm signal to both fans.
Your pump is not pwm, it's analog. It has 12v+, G, tach. Your fans are pwm. If you put the fans on the 4wire side, the rpm of the fans gets read on the tach and changed by pwm. The pump has zero affect or monitor since there is no tach for it on the 3 wire side of the splitter, the connection kills the tach feed. It's the fan that gets monitored.
If you put the pump on the 4wire side, to use the tach to monitor pump rpm, it kills the pwm signal, there's nothing to control the fans as the tach doesn't change under a pwm header. If you change the header to analog/voltage then the pump rpm can be changed by voltage, but that also changes the fan rpm by voltage too.
A header only reads 1 input, only controls 1 output, that being whatever the pin location points to. If the pins point to a fan, everything pwm on the splitter is controlled and monitored as if it was that particular fan. An analog pump as control/monitor has zero ability for control as it has no pwm. So any pwm fans on the splitter whether it's a 2,3 or 4way splitter with ganged splitters afterwards, are stuck at 12v+ constant.