Improve Redis memory management

[tokebe]

Our Test instance, likely due to a high level of traffic, caused the redis pod to go out-of-memory and requiring a flush of the entire redis store. This prompts discussion of how to avoid this in the future.

Currently, we store results for 30 days. We may wish to reduce this time, or actively check redis memory usage and drop the oldest results to bring it back to some threshold. If we do something more complex than shortening the result storage time, we may need to revamp the system to store some sort of index of results, as we currently don't do so and may need it to reason about what's stored, how much space it's taking up, and its age.