Currently, if an exception occurs in the sync client event loop, the process is terminated with very little - if any - information about what went wrong. Instead, the event loop should be structured in a way where operations being executed are tied to a Realm/session and any exceptions are logged, reported as fatal errors through the session error handler, after which sync stops for this realm/session, but remains active for any other sessions.

A very common example of the current broken behavior is when multiple processes try to open the same Realm file and get MultipleSyncAgents error. When this happens on Windows, the process crashes with no information about the cause of the problem or how to fix it.

This is the second issue that manifested in this bug report: https://github.com/realm/realm-dotnet/issues/3437.