Interesting that the reduced T wallkick chart Koryan made is finally explained. All pieces that can fit in a 3x3 square apparently have the same wallkick offsets, and if you extrapolate this pattern to the T piece you get some wallkicks that are impossible to trigger. An observation: In the very first I example, wallkick 2 and 3 could be switched without a functional difference.
Which saves about 20 bytes in LJ0.11 source code. But wallkicks 2 and 3 in "I2 from" and "I4 from" seem to behave very asymmetrically. I can't see any way this would be on purpose. When I get time I'll check most of the I stuff against TDS.
ahh yeah i forgot to mention it... we can't deduce the priority order on 4 cases between WK 2 and 3. In fact, each time you want to switch from vertical to horizontal orientation because it's impossible to create a conflict.