Well, from the 100,000 simulations I ran, the average kills before the "Champion" falls was 7.90 with a standard deviation of 2.84 kills. A maximum of 22 kills (9 times) and 26 times with no kills at all.
I don't know how you handled initiative, but I gave the Champion a +1 modifier and the 1st-level fighters no modifier, with the Champion also going first in the case of ties. I also gave the Champion the 103 hp average and randomly added the second wind if hp dropped below 82 (since his max boost would add 21, returning him to full). The action surge was added the first round. Like you, replacements moved in when kills happened so the Champion was always attacked 8 times per round.
The chart shows the observed number of times X kills occurred. Your calculated estimated kills of 10 is high a bit IMO and I hope this helps. If you want the excel spreadsheet with the macro, just let me know.
View attachment 107119
I figured someone would piggy back off the work and take it to the next level...kudos! My guess is I underestimated the second wind power for the 1st level fighters, so more of them died than should have.