Tests how console messages from worker get into page's console once worker is destroyed.

=== console.log event won't get lost despite worker.terminate. ===
Starting worker
-> Posting to worker: message0 (posted after starting worker)
Terminating worker
<- Log from page: message0 (posted after starting worker)

=== Scenario with autoattach enabled and stopped. ===
Starting worker
-> Posting to worker: message1
Starting autoattach
child session for worker created
Sending Runtime.enable to worker
-> Posting to worker: message2 (posted after runtime enabled)
-> Posting to worker: throw1 (posted after runtime enabled; yields exception in worker)
Stopping autoattach
-> Posting to worker: message3 (posted after auto-attach)
Terminating worker
<- Log from page: message1
<- Log from page: message2 (posted after runtime enabled)
<- Log from page: Uncaught (in promise) throw1 (posted after runtime enabled; yields exception in worker)
<- Log from page: message3 (posted after auto-attach)
<-- Console API from worker: message1
<-- Console API from worker: message2 (posted after runtime enabled)

=== Scenario with autoattach from the get-go. ===
Starting autoattach
Starting worker
child session for worker created
-> Posting to worker: message4 (posted before worker's runtime agent enabled)
Sending Runtime.enable to worker
-> Posting to worker: message5 (posted after worker's runtime agent enabled)
Terminating worker
<- Log from page: message4 (posted before worker's runtime agent enabled)
<- Log from page: message5 (posted after worker's runtime agent enabled)
<-- Console API from worker: message4 (posted before worker's runtime agent enabled)
<-- Console API from worker: message5 (posted after worker's runtime agent enabled)

=== New worker, with auto-attach still enabled. ===
Starting worker
child session for worker created
-> Posting to worker: message6 (posted just before worker termination)
Terminating worker
Stopping autoattach
<- Log from page: message6 (posted just before worker termination)