feat: assert no error event during ready -> stale -> ready #387

Context

Part of open-feature/flagd#2027.

The ready -> stale -> ready scenario in gherkin/events.feature registers an error event handler but never asserts it stays silent. So a spurious PROVIDER_ERROR during a short (< grace period) disconnect goes uncaught. flagd#2027 changes the default retryBackoffMaxMs/retryGracePeriod to 5s/10s so a normal reconnect resolves inside the grace window (STALE -> READY, no ERROR); this scenario should assert exactly that.

Scenario change needed

• In the ready -> stale -> ready scenario, add a step asserting the error event handler did not fire:
  And the error event handler should not have been executed

Notes

• This introduces a new gherkin step; provider step definitions must implement the negative assertion when they bump to a testbed release containing this change.

Related

• Default retryGracePeriod and retryBackoffMaxMs are pathological flagd#2027 (parent)
• feat: assert no error event during ready -> stale -> ready #386 (PR)