Peter,
Here it is, Fig 12 at M2.8
That figure does indeed show the TOTAL bleed to be around 1/3 intake capture, but the bit we are interested in is the shock trap bleed which becomes the cooling air and the eventual secondary nozzle air. That bleed is 5% of the intake capture, but since at that point the engine flow is only 75% of the intake capture the secondary airflow would have been 6.7% engine flow. Less than my estimate for M 3.0 from Peter Law's data but reasonably close all things considered.
The following is a previous post of mine. Is there any merit in my "schoolboy" plot and conclusions? I know it's just observed FF and not sfc, etc but was I just lucky or have I misinterpreted things? The range charts I got from his book.
There is something like that going on, although one needs to be careful what data one plots. At constant altitude and weight for example the fuel flow increases steadily with increasing Mach No up to Mach 3 then rises much less up to M 3.15 then resumes its inexorable increase. But the specific range is better at M 3.15 than at either M 3.0 or M 3.2.
It certainly looks as if the powerplant gets close to its maximum efficiency around Mach 3.1~3.15 probably, as you say, because it gets to maximum capture (shock on lip) conditions and the engine can swallow enough air to eliminate the need for any forward bypass bleed. Maybe some of the other references you have posted will show up why.....