Calculating for efficiency is more useful than "shots per x,y,z". Similar to how MPG is more a useful a stat than gallons consumed on a trip of x miles.
Graphing efficiency as you empty a tank is significantly more useful than just obtaining the average efficiency for the entire tank range. Most modifications do not merely offset the graph up/down. They tend to increase performance at one pressure while decreasing performance at another.
Measuring mass favors larger tanks while measuring pressure favors smaller tanks. Measuring mass is also more cumbersome, it takes significantly longer to graph efficiency and makes graphing efficiency while playing impossible. Consider a 13ci tank while measuring pressure changes. Each shot drops tank pressure a significant amount compared to a 68ci. Therefore a less critical gauge/sensor can be used. Fewer shots need fired as well. A very accurate mass system prefers a large tank which therefore requires more shots.
I would argue the compressibility has very little impact on any efficiency or fill/volume/pressure correlation calculations. In the above examples, the mag's system efficiency change was an order of magnitude greater than the compressability. This isn't a scientific experiment. We are simply doing some back of the napkin stuff using measurements that are easily +/-5% and operating under the assumption that our results will easily vary that amount.
Paint is adequate because of what we are doing and how paint variations relate to the energy equation.
We are looking at maximum theoretical energy potential and comparing that to useful energy out. Any energy lost in the gun's operating system, lost to friction or turbulent air, energy imparted to spinning the ball, etc, etc,.. Doesn't matter, it's all put in the same category, inefficiency.
Paint ball mass and velocity are the only two things that relate to 'useful energy out'. Velocity is easily measured. So if our gun system has a hard time with lumpy paint and fires it at a lower velocity.. well that shows up as a less efficient system because that's what it is. A gun system that can fire lumpy paint without a reduction in velocity will show up as a more efficient system.
Regarding irregular paint mass...
The calculation useful energy out is E=0.5mv^2. Since m has a power of only 1, we do not need to measure the mass of each ball. The average paint ball mass is adequate even if there's a large ball-ball variation. Try it for yourself with the following two data sets.
3g 100fps, 3g 200fps, 3g 300fps, 3g 400fps
2g 100fps, 4g 200fps, 1g 300fps, 5g 400fps
You should notice equal total energy even though the mass fluctuates in the bottom set.