I'm having the same problem. My acq_params.txt is only 2 lines (blipup, blipdown), so my index.txt is just 1 1 1... 2 2 2..., and it works when no --resamp is specified (but it outputs both sets of images). However, when I add --resamp=lsr I get the error "eddy: msg=These data do not support least-squares resampling".
Jesper's advice was to "Make sure that you have duplicated all diffusion directions with opposing PE-directions". I verified that the diffusion directions were identical (i.e., duplicates) for the two opposing PE-directions. Are there additional manipulations to the bvecs file to be made?
One note that may or may not be relevant is that I have 3 b-values (b=0, 1000, 2000). These are also duplicated in the bvals file, and it's not a problem without --resamp, but there's nothing else strange I can see.