I think it would be best if you avoided using the current q2-dada2 for pacbio longreads even though they may appear ok. The dada2 version in QIIME 2 is currently at 1.10 and some important changes were made to handle long reads at 1.12 version (which the tutorial you've seen uses). It's possible that by the next release of QIIME 2 this will become available (see issue here) but that's not a certainty either. Your best bet would be to use the R version of DADA2 at the moment.