Long running queries run long after the graphql request has timed out

[proehlen]

More of an issue while data is missing from the db but still an issue:

If the user makes some request that results in a very long running postgresql query, the request will timeout with an error but postgres will keep trucking for minutes or hours trying to answer the query. This is a potential DDoS vector as well as being inefficient.

Need to set postgres timeout option after the migrations have run per brianc/node-postgres#518.