fix: handle async call not being able to be called

[matt2e]

Fixes case where async call's early failure skipped clean up and retries.

Issue steps:

• Deploy a module
  • Deployment and related rows get added to db, but no runner has been assigned yet
• Add an async call that calls that module
  • A controller picks up the async call and attempts it by calling callWithRequest(...)
  • This gets the schema by calling getActiveSchema which gets the schema filtering by minReplicas > 0 and r.assigned = 'assigned' (link)
  • callWithRequest(...) fails because this new module is not in the fetched schema
  • executeAsyncCalls(...) return early due to this error, not updating the state of the async call
    • the lease gets released rather than timing out (due to the defer in executeAsyncCalls(...))
    • we end up with an async call with state = executing and no lease

So this async call gets stuck forever. For pubsub this means the subscription stops forever.