Hi @emmlemore, welcome to :qiime2:!
Correct. Note: the files towards the top of that page are pre-trained classifier artifacts for the full-length SSU data, and the V4 region. If you'd like to create the classifier yourself, then you'd use the other files further down the page. Again, you do not need to import these as they are already QIIME 2 artifacts.
You do not need to use these parameters. But they are helpful in removing sequences that might be too long or too short for your needs. I often do not use the --p-trunc-len unless I am dealing with highly variable sequences, e.g. ITS, or if I am only using R1 (forward) reads. Remember, if a sequence is shorter than the --p-trunc-len value, it will be discarded, longer reads will be trimmed to this length. For the other two parameters, if I use them, I'll often add / subtract 10-30% of my expected amplicon length so I have some breathing room. That is, I might set the following: --p-min-length 180 --p-max-length 280.
Yep!
Yep!
Just an FYI, we have a very detailed tutorial for you here.