You had interpreted what should have been the appropriate results for both cases, but there were "enhancements" to the code. The first item concerning missing child output should already be corrected in the already released code, so the latest version should have that update. The second item will be corrected in the next release following this posting. A work-around is to use the -save_vr option to get the motion parameters for all epi datasets.
Thanks for the feedback.