evan9162 - I do believe that you have a proper setup for testing. Q5 and P4 is way far apart. I have different batches of Q5s they are quite close in terms of light output. I do have a Q5 which is a fluke, after a 2.5A attemp
For others, proper testing of the vf and current requires a semi-permanent testbed. There are certain variables that may or can affect your readings such as die temperature, ambient temp. Some metal oxidize over time, make sure that you use silver plated connectors. If you are using sense resistors/shunt for current reading, it's best to solder the reading points permanently. You will also need something that can read the temperature of the die. It doesn't need to be the accurate real die temperature but note the location the reading was taken. The precision of your equipment is also important. Have you done a 1yr calibration lately? Finally let the CREE, power supply, and multimeter warm up to a stable temperature before tests.
As I said, from my experience the Cree from the same batch are consistent. I have to add that they do differ from batches to batches. I know I have a reliable source, the cree Q5s I have was removed from sealed reel. I marked them, I use them as a 'benchmark' control to test other Crees.
If you have a Cree Q5 that outperforms the others, you can make it your 'real' Q5. If new Q5s performance that you have is not even close to your real Q5, then there is something to suspect. Could be a problem with your source.