closes #105

The test is checking for notes, but 'Bridge Fed' is, in fact, an article according to the post discovery algorithm. Also, I'm adding as note that the main difference between a note and an article is that the latter has a title, while the first doesn't.

* looks through the full mf2 tree for step 7 of authorship https://github.com/indieweb/authorship/issues/2
* if no author h-card is found, falls back to returning the author URL without other data instead of missing author

closes #95

if the document contains a link to the target, but that link is not in the parsed result, don't return an error, but also don't return the parsed document

closes #150

XRay now looks for images inside the HTML and does not consider those empty posts

To make it possible to consume feeds of pre-rendered Microformats2 JSON
data, we need to be able to discover that it is a Microformats2 JSON
file using the known MIME types documented  in
http://microformats.org/wiki/microformats2-mime-type

for https://github.com/aaronpk/Aperture/issues/56

closes #89

closes #92

closes #86

more consistently returns HTTP 410 now

This adds a new property, "meta", which includes alt text and has room to include additional properties later.

closes #85

replaces #78

closes #79

* a single h-entry and h-card, where the h-entry has no URL, will result in a permalink page with that h-entry
* multiple h-entrys followed by an h-card is a feed

closes #77

including from rel=alternate

closes #73

parse from both link tags and the u-redirect-uri property

adds script to refresh the downloaded instagram data for the tests as well

instagram seems to have rolled out the `graphql` key everywhere now

main difference is the deprecated rel handling

closes #62

also expands parsing to be able to handle twitter JSON from the streaming API which is subtly different from the HTTP API.

closes #61