Divide up discovery techniques by document or URL

Discovery from HTML can have two inputs:

An URL of an HTML document, like https://html.example/some/path/image-14.html.
A document, like <!DOCTYPE html><html><head><title>Example</title></head><body><img src="image-14.jpg"></body></html>

The discovery techniques are different for each input. If discovery by URL does not work, the user can try discovery by document, and the inverse.

Similarly, discovery from ActivityPub can have two kinds of inputs:

An URL or ID of the ActivityPub object, like https://ap.example/objects/person-22.jsonld.
A JSON-LD document, like {"@context": "https://www.w3.org/ns/activitystreams", "id": "https://ap.example/objects/person-22.jsonld", "type": "Person", "name": "Person Twenty-two"}

The discovery techniques are different for each.

swicg / activitypub-html-discovery