Support as many content address formats as possible

[btrask]

Related to #110, we should strive to support all of the different content address formats:

• Our own hash URIs, of course
• RCF 6920 (ni:)
• Algorithm-prefixed hashes (e.g. as used by Subresource Integrity)

There is no reason we can't resolve these hashes and even generate them in our UI.

Unfortunately, supporting the addresses used by BitTorrent, Camlistore and IPFS (https://github.com/ipfs/go-ipfs/issues/1953) is nearly impossible, because each one generates its own unique hashes, which cannot even always be consistently derived without outside data.