I've got a case where there's a crasher in the seed corpus (because we're using another project's seed corpus) -- but ClusterFuzz doesn't appear to be finding that, even though just starting the fuzzer with the corpus in my local environment triggers it reliably.

Is this intentional?