JoeJ said:
The leg in air is irrelevant for the other leg, so we can ignore it completely.

Sorry, I have to disagree. The leg in the air is actually what's driving the entire animation. That leg has to arrive at a specific spot on the ground, and it drives the knee, which drives the thigh, which drives the pelvis, which drives the other thigh, all the way back to the foot which is fixed

You can use the same IK mechanism for the other leg in air too. I did this experiment: Stand with one foot on that moving platform, while keeping the other foot fixed at a target point in air. The only connection between two two limbs is the pelvis. But you can move that too, freely, while the feet keep at their individual targets. So if the air leg has to drive pelvis, i don't see why that's a problem affecting the stance leg. IK will handle it, that's what it's good for. Ofc. you need to make sure targets are in reach, but the rest should just work if done correctly.
Really the only problem i see is: It won't look like natural behavior in many cases, mostly because neither animation nor IK care about balance. Also if velocities are constant, it looks like primitive robots. It looks much better if you try to keep acceleration constant, but that's another topic.

… oh, and if the problem becomes complex (e.g. multiple objectives like different targets for multiple limbs), some iterations to minimize error are necessary in theory. But i never had to do this. Because timestep is small, error is small too and it converges quickly.