Cache is not fully cleared for the timeout event.

[IvanChupin]

For the org.openkilda.wfm.topology.switchmanager.service.impl.SpeakerWorkerService

There might be a case of timeout event so that handleTimeout() method is called.
And inside this method keyToRequest map is cleared properly, but chunkedMessageIdsPerRequest cache is not cleared.
So it might be a case where we are handling the chunked responses and in a half way though we might not fully receive the chunked chain of messages. In that case we expect to receive timeout event. This should lead to handleTimeout() method call, which should also clear chunkedMessageIdsPerRequest map.

Implement proper cache(chunkedMessageIdsPerRequest) removal for appropriate request key.