Hmm, interesting. Not sure what to make of all of that but it did prompt me to play around with CPU scaling.
If I put the CPUs into performance mode (echo performance > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor, same for cpu1), fq_codel/simple can keep up with line speed. If the CPUs are in ondemand mode, it cannot.
Interestingly, during a DSLreports download test, in ondemand mode, the CPUs don't both top out at top speed, the highest they get is:
root@LEDE:~# cat /sys/devices/system/cpu/cpu*/cpufreq/scaling_cur_freq
1000000
1725000
so it seems the issue isn't raw CPU power, it's something else (perhaps synchronization latency, which I took as a hint from reading the referred threads).