Hi,
I'm not part of the QIIME2 team, but I am curious as to why you want to use fasta files. Do you also have the quality score files that go with the fasta files? If so, it might be more efficient to convert the fasta and quality score files into fastq files and then import. Also, how do you plan to generate a Features Table? I am less familiar with Deblur, but I know DADA2 uses quality scores as part of the denoising process. And I am pretty certain Deblur uses the quality scores to define an upper bound on error rates. This means that you probably can't generate a feature table from a fasta file using either DADA2 or Deblur.
This paper compares DADA2, Deblur, and UNOISE3: https://peerj.com/preprints/26566.pdf Maybe it will be of some use to you?