Sometimes we will generate a lot of errors in a small space of time. Raygun is very good at aggregating these errors but I don't really need 10K+ examples of an error to be able to diagnose it.
When high frequency errors occur they eat up the message allowance very quickly.
Ideally these errors would be throttled by the raygun client and only the details of a sample set sent to raygun. I would however still like the total number of errors in the group so I can clearly see that there is a large problem. The errors whose details are dropped at the client should not count towards my usage rate or maybe billed at a lower rate.