Great investigative work! Even though the set up you described is a very ideal starting spot, it isn't always so in all rooms - as you have seen. We can achieve pretty good results in our test room the long way as well. Additionally, low frequencies and high frequencies can behave very differently in each room. Facing the long way for instance in a decent size room makes first reflections, even without any treatment, very low in volume and so not as much as a problem. Around 100-200 Hz it doesn't look like either way has a clear winner. But above and below that, facing the long way does seem to be better.
I will offer one caveat though - suggestions are just suggestions for starting points. It is possible that optimizing position in each direction may actually come out with a much clearer winner. But I like the low frequency response of the long wall better, so personally, that's what I'd try to optimize. Since you're already versed in measurement by now, I'd say give it a go one weekend. Try changing each of the variables independently and running tests to see how they affect your response - back a foot, forward a foot, wider by a foot...moving seat position back and fourth, etc. It's a great experience since you'll be able to visually see exactly where some of the peaks and valleys are from, and how to change what frequencies are boosted and cancelled in some respects.
The actual waterfalls of the room look great though in the long wall measurement given your somewhat modest corner treatment; about 40dB of decay within 300ms across the board is really quite good. You could get less noise in your measurements by using multiple sweeps per measurement, or longer sweeps, or both. I usually do 4 sweeps @ 256k, which is good enough for most purposes.
