Improve error handling of existing messages fetch and never retry

There are at least two user reports that fetching existing messages
sometimes results in infinite loop of retrying it. Account is working
if set up from the backup, but never starts working if set up
from scratch.

This change improves error reporting, but also sets FetchedExistingMsgs
before actually trying to do it. This way if the operation fails,
connection is reestablished, but fetching existing messages is not
retried again over and over.

CHANGELOG.md

@@ -21,6 +21,7 @@
 - Fix uncaught exception in JSON-RPC tests #3884
 - Fix STARTTLS connection and add a test for it #3907
 - Trigger reconnection when failing to fetch existing messages #3911
+- Do not retry fetching existing messages after failure, prevents infinite reconnection loop #3913
 
 
 ## 1.104.0