Refactor to define document types via configuration?

[jimallman]

@jar398 and I have talked about a general refactoring that would potentially remove lots of boilerplate code by defining document types (studies, collections, amendments, illustrations, etc.) with configuration. Sort of a type registry where we define:

• Pattern (regexp) for defining a valid document ID
• Shard naming convention
• Distinguish “folderish” types (eg, study, illustration) from simple JSON documents
• Limitations on file sizes, file types
• Pattern for shard folder structure and singletons (eg next_ott_id.json)
• Patterns for corresponding public URLs for a document

I'm pretty sure there would still be a fair amount of type-specific code for validation, testing, etc. But this might drastically shorten the time and expertise requireed to add new document types.