Can you post your output log from afni_proc.py?
You should check to make sure the slice pattern is included in the NIFTI files as Siemens scanners often don't encode that in DICOMs.
3info -VERB 6.120.1+orig.HEAD
If you don't see any slice timing information in the info, and the output of afni_proc.py script says "already aligned in time", you should specify a slice timing in the script. Siemens usually defaults to that for an even number of slices acquisition, so you can use something like this:
-tshift_opts_ts -tpattern alt+z2
In terms of censoring: I think it makes more sense to censor motion to 0 than to include a value that is going to be often outlandish and unrelated to your stimulus. If you feel extra enterprising, you could go back and interpolate those values as the average of the time points around it. But realistically you shouldn't have that many time points being censored and it should have little impact on your ISC correlation.
Finally, it's worth noting that a single post on the message board would likely be sufficient. I see that Paul and I are now both answering different threads of your question.