Improved Sumo summary data fetching (#14551)

Sumo data is loaded so the application stays responsive while it waits.

Run Sumo transfers on a dedicated thread
Split RiaSumoConnector into transport and data delegates
Load Sumo summary vectors concurrently and without waiting
Show that Sumo data is being loaded

ApplicationLibCode/Application/Tools/Cloud/RiaSumoDefines.h

@@ -22,6 +22,8 @@
 
 #include <QString>
 
+#include <cstddef>
+
 using SumoAssetId = nonstd::ordered<QString, struct sumo_asset_tag>;
 using SumoCaseId  = nonstd::ordered<QString, struct sumo_case_tag>;
 
@@ -29,4 +31,9 @@ namespace RiaSumoDefines
 {
 QString tokenPath();
 int     requestTimeoutMillis();
+
+// The timeout of a request nothing is waiting for. Only there so a request that never answers is eventually
+// given up on, and generous because a summary vector that has not been aggregated yet is produced on demand
+// by the request asking for it. Nothing is blocked while it runs, so waiting longer costs nothing.
+int asyncRequestTimeoutMillis();
 }; // namespace RiaSumoDefines