Conventions for annotation_rules?

[bmcfee]

Since #1 is putting us on the path toward re-importing all the existing datasets, we may want to settle on some conventions for the annotation_rules field in AnnotationMetadata.

I'm mainly thinking of tagging data, where there are a few modes of annotation that are common (strong vs weak, positive-only, etc.) Maybe other tasks also have a small set of standard conventions that we can list out here and use in the parse scripts?

[justinsalamon]

By all means. At the same time, there are datasets (e.g. MedleyDB) that have quite an involved annotation process (use this tool, follow these steps, etc.) - which (in an ideal world) should also go in that field.