as long as the same equipment is used to inject the air the data coming out the other end will be comparable between tested mufflers. Might not be all that comparable to an installation, since number and type of bends and size pipe comes into play, but will at least give the POTENTIAL flow of the muffler, or max flow under ideal conditions. That's SOMETHING to go on!
I was thinking also... if you were to put an electronic tone into the collector end, and a microphone on each port tube, on an oscilloscope, then swept the frequency through the equiv. RPM range, you could measure tube length (for equal length and bends that affect virtual length) I bet you could find the optimum scavenge RPM within a few percent.
When the tubes are all series-resonant the load on the transducer on the collector end would go up. You could actually measure header performance.