During pruning of a LARGE quantity of alerts we discovered the process was not completing in the allotted timeout.  in 4.6 we updated pruning to continue on error.  Meaning if a batch failed we would process the next batch instead of returning and error and rolling back the transaction.  

However, if a given batch fails because the context is cancelled, all subsequent batches will fail because the context will remain timed out.  We should at a minimum detect the context timeout and process no further batches.  We may want to take it a step further and also check for transient postgres errors before continuing.  

https://github.com/stackrox/stackrox/blob/285e8e1efa6b4beb56e862c5f5c1e4f04d47021f/pkg/search/postgres/store.go#L538