Dynamic metadata resolution and transformed identifiers

[Cantor, Scott]

The slightly problematic thing is that you're coupling the transform used
on the server-side for implementation reasons with the one used in the
client.

The {SHA1} thing is a special case since it had to be made interoperable
for artifact binding use, so it's required to implement.

In practice, it takes not much more than a simple bash script to take a
URL-encoded entityID parameter in the URL and decode-then-hash it into the
filename to return. I've done it, in fact.

-- Scott