Doing a bit of listening yesterday and today I have found that the LS50 Metas which I used at the show sound considerably better without the KC62 subs in circuit, and with the sponges removed from the rear port holes.
I bought the speakers shortly before the show and didn't have a chance to test this. I was tempted to do this at the show, but I decided not to change anything once it was sounding adequate, for sanity. A lesson for next time: spend more time optimising, without preconceptions.
This is a bit controversial, but I can't help feeling that a likely explanation for this is that the AD-DSP-DA stage in the KC62 subs is not actually transparent. It has become a professional dogma that AD-DA is inaudible, but somehow the straight analogue signal path seems to sound fresher and more real in the higher frequencies when the subs are out of circuit. Maybe it's the one extra interconnect, but somehow I doubt that.
I intend to use a different speaker set-up at the next show where I have a room. The LS50 Metas sound great on their own, and the KC62s do a great job for their size, but I won't be feeding the LS50s from the KC62 high-pass throughput again.
The next public SUPATRAC demonstration will be better.