Yep, really anything that the types let you do should be more or less fine for any amplicon. I think the only thing you should keep in mind is the merging of your reads (if paired) and whether you have enough length in your sequences to make that work.
Otherwise, I believe the Silva full-length classifier should work great. If this is common enough for you, you could also train a classifier specifically on the V1-V2 region, but for any one-off work, the full length is fine.