ODL: Discovery document

[HadrienGardeur]

To facilitate the discovery of ODL sets/catalogs/feeds (we'll need to decide how we call them), we could extend the ODL draft (https://drafts.opds.io/odl-1.0) to include a discovery document as well.

For this document, I've identified the following requirements:

• it SHOULD use the same RWPM/OPDS 2.0 syntax
• it SHOULD allow a client to discover URI templates where content can be harvested from
• it SHOULD contain metadata that could be presented to either a human or a machine to decide what should be harvested
• it MAY indicate that a given URI template will require authentication

I'll maintain an up-to-date example of how this could be handled at: https://gist.github.com/HadrienGardeur/2599ae6b33d1e87caaee53434937e546

• We would define a new collection role. I'm not sure what we should call it but I've used harvest in this example.
• We can re-use metadata to describe each collection that can be harvested, in this example there's only title but description remains relevant as well.
• We might include additional metadata over time to provide useful information about the nature of these collections.
• Each collection contains a URI template where a client could harvest content from.
• We'll need to define a list of parameters, for now I've added since and until to facilitate partial requests over a specific date range.
• Do we need a specific rel for that link? This could be useful if for some reason we end with multiple links.
• Do we need to define new media type for ODL as expressed in OPDS 1.0 vs OPDS 2.0 ? This could be useful if a content provider has both.

cc @pbrantley @jce1028 @jonathangreen