Add support for large dataset results

[bradgessler]

Queries with very large results can timeout Datagram. Run the queries in a background job and store them in a file or object store.

Issue #3 is relevant for tracking a query that's "in progress" via versioning. If the results are stores in a file or object store the versioning mechanism could easily be used to deal with large async queries.

Currently we're just sticking a LIMIT 100; at the end of a query if a LIMIT isn't specified. Ideally we'd do this pagination transparently/lazily if you run a query from the web view and just want to see the results (e.g. load query results in 100 row chunks).

For larger/longer queries the server just times out.