Caching

How should resolution results be cached?

This could be influenced by:
- Default settings by community consensus and recommendations of the IND WG.
- Custom configuration by the UR administrator.
- Request parameters by a UR user (e.g. to override defaults and force a fresh non-cached lookup).
- Settings associated with the identifier (e.g. TTL setting in the DID document or BNS zone file).

How should the cache be implemented? The UR could cache results in memory, or use a shared caching mechanism such as Blockstack's [Atlas peer network](https://github.com/blockstack/atlas).