pcbnew: Parallelize initial board triangulation

Triangulation of zones can be time consuming for large boards.  This
adds a std::thread implementation of parallel zone loading to speed this
process on multi-core machines.

(cherry picked from commit a04ef37d67)

pcbnew/pcb_draw_panel_gal.cpp

@@ -140,13 +140,25 @@ PCB_DRAW_PANEL_GAL::~PCB_DRAW_PANEL_GAL()
 
 void PCB_DRAW_PANEL_GAL::DisplayBoard( BOARD* aBoard )
 {
+
     m_view->Clear();
 
-    // Load zones
+    auto zones = aBoard->Zones();
-    for( auto zone : aBoard->Zones() )
+    std::atomic<size_t> next( 0 );
+    std::atomic<size_t> count_done( 0 );
+    size_t parallelThreadCount = std::max<size_t>( std::thread::hardware_concurrency(), 2 );
+
+    for( size_t ii = 0; ii < parallelThreadCount; ++ii )
     {
-        zone->CacheTriangulation();
+        std::thread t = std::thread( [ &count_done, &next, &zones ]( )
-        m_view->Add( zone );
+        {
+            for( size_t i = next.fetch_add( 1 ); i < zones.size(); i = next.fetch_add( 1 ) )
+                zones[i]->CacheTriangulation();
+
+            count_done++;
+        } );
+
+        t.detach();
     }
 
     // Load drawings
@@ -171,6 +183,14 @@ void PCB_DRAW_PANEL_GAL::DisplayBoard( BOARD* aBoard )
         m_view->Add( aBoard->GetMARKER( marker_idx ) );
     }
 
+    // Finalize the triangulation threads
+    while( count_done < parallelThreadCount )
+        std::this_thread::sleep_for( std::chrono::milliseconds( 10 ) );
+
+    // Load zones
+    for( auto zone : aBoard->Zones() )
+        m_view->Add( zone );
+
     // Ratsnest
     m_ratsnest.reset( new KIGFX::RATSNEST_VIEWITEM( aBoard->GetConnectivity() ) );
     m_view->Add( m_ratsnest.get() );