Handle badchangeset error when printing changeset contents in debug

[michael-wb]

What, How & Why?

After the updates from the unify core error handling project, the realm-sync-tests were crashing due to an "unknown instruction type" if debug level trace or higher was enabled. This was due to an exception being thrown by parse_changeset() during the Sync_BadChangeset test, which intentionally creates a bad changeset.

Updated debug output handling to catch the BadChangesetError exception and print a message instead of passing this exception up.

Also, fixed the race condition in the Sync_DetectSchemaMismatch_* tests that was resulting in the realm::TableNameInUse: The specified table name is already in use error. This was due to the first session completing sync before the write transaction transaction on the second session was completed. Moved the call to nonsync_transact_nofity() until after both write transactions were completed.

Also, updated the ssh connection timeout to 60 secs for scp and ssh operations when starting remote baas, since the startup of the remote baas was sometimes failing out. This was likely due to the original connection timeout of 10 seconds.

☑️ ToDos

• 📝 Changelog update
• [ ] 🚦 Tests (or not relevant)
• [ ] C-API, if public C++ API changed.

[danieltabacaru]

What's with the evergreen failures?

[ironage]

Nice 👍