The network error checks in the EventSourceImpl::OnStopRequest might not be exhaustive.
We need to investigate if there are additional network errors which need to be added in the
EventSourceImpl::OnStopRequest.

Our current approach is we attempt to reconnect only if the errors match a set of errors.
As per the spec we need to re-establish connections for any errors except when the user has aborted the request or the user agent knows that the network request attempt is going to be futile. Therefore, we need to make an attempt to re-connect to the server in all cases except for user initiated abort or permanent network failures. This must be our approach for a long-lasting solution.

We need to also add tests to verify 1809332, 1706003 when we implement this refactoring.