If it counts the pulses properly then I guess you're doing it right. However, doing it this way is probably not the best when you're trying to get more sensors in but more on that later.

If you look at the product page for the sensor there's a diagram showing the output frequency as a function of the flowrate. The output seems pretty linear (as you'd expect) with 4.5Hz/l/min.
So if you count pulses for on second and divide the result by 4.5 you have your flowrate in litres/minute.

For multiple sensors I'd set up a timer interrupt and poll all inputs in one go. Compare this state to the previous one to see if any input changed. Since the dutycycle of sensor output is 50% you can get twice the resolution (if needed) by counting both the positive and negativ going edges of the signal but then you'd need to poll the inputs twice as fast to guarantee not missing any pulses. It's going to be kind of like reading incremental encoders (lots of examples floating around here) but without the need to determine direction.

Nice hardware setup!

/Henrik.