fix(physics): the external program messages (multiplayer) no longer keep the physics steps in one thread, only network sessions do; the debug panel tells what keeps them in one

rendering/opengl46/opengl46renderer.cpp

@@ -866,7 +866,7 @@ void opengl46_renderer::SwapBuffers()
                 ( counts.threads > 1 ?
                     "steps on " + std::to_string( counts.threads ) + " threads: " + std::to_string( counts.groups ) + " groups, largest "
                         + std::to_string( counts.largestgroup ) + ", " + std::to_string( counts.maingroups ) + " in the main thread (forces by kind not timed)" :
-                    std::string{ "steps on 1 thread" } ),
+                    "steps on 1 thread (pool of " + std::to_string( counts.pool ) + ( counts.singlethread.empty() ? std::string{ ")" } : ", off by " + counts.singlethread + ")" ) ),
                 "" );
             char const *const kindnames[] = { "rail, within ", "rail, farther", "road, within ", "road, farther" };
             text += timeline( 3, "by kind (near: " + nearlimit + ")", " count  stand  forces  update" );

utilities/Timer.h

@@ -204,6 +204,8 @@ struct subsystem_stopwatches {
         // the steps of the physics: threads, the groups of vehicles they were split in, the largest one, the groups done in the main
         // thread (vehicle_table::update_steps_parallel())
         unsigned int threads { 1 };
+        unsigned int pool { 1 }; // threads of worker_pool
+        std::string singlethread; // what keeps the steps in one thread, if anything
         unsigned int groups { 0 };
         unsigned int largestgroup { 0 };
         unsigned int maingroups { 0 };

vehicle/DynObj.cpp

@@ -8681,15 +8681,22 @@ vehicle_table::update( double Deltatime, int Iterationcount ) {
     Timer::subsystem.vehicles_prepare.stop();
     Timer::subsystem.vehicles_forces.start();
     auto const totaltime { Deltatime * Iterationcount }; // całkowity czas
-    // the steps of the physics on many threads (Global.performance_cpucores). not in a network session, which keeps the draws of
+    // the steps of the physics on many threads (Global.performance_cpucores). not in a network session (network.server, .client),
-    // Random() in step across the machines: the threads have engines of their own
+    // which keeps the draws of Random() in step across the machines: the threads have engines of their own. the messages to
+    // external programs (Global.iMultiplayer) don't mind
     auto &workers { Workers() };
     auto const parallel {
         ( workers.size() > 1 )
-        && ( Global.iMultiplayer == 0 )
         && Global.network_servers.empty()
         && ( false == Global.network_client.has_value() ) };
     counts.threads = ( parallel ? static_cast<unsigned int>( workers.size() ) : 1u );
+    // why not, for the debug panel
+    counts.pool = static_cast<unsigned int>( workers.size() );
+    counts.singlethread =
+        ( workers.size() <= 1 ? "performance.cpucores " + std::to_string( Global.performance_cpucores ) :
+          false == Global.network_servers.empty() ? std::string{ "network.server" } :
+          Global.network_client.has_value() ? std::string{ "network.client" } :
+          std::string{} );
     if( parallel ) {
         update_steps_parallel( Deltatime, Iterationcount, coarse );
     }