Fix management of mRunScheduled state in reporting engine. (#20694)

Without this change, we can schedule a reporting run, then do a sync
Run (due to urgent events), then (unnecessarily) schedule another
reporting run.

src/app/reporting/Engine.cpp

@@ -539,6 +539,7 @@ CHIP_ERROR Engine::BuildAndSendSingleReportData(ReadHandler * apReadHandler)
 void Engine::Run(System::Layer * aSystemLayer, void * apAppState)
 {
     Engine * const pEngine = reinterpret_cast<Engine *>(apAppState);
+    pEngine->mRunScheduled = false;
     pEngine->Run();
 }
 
@@ -575,8 +576,6 @@ void Engine::Run()
 
     InteractionModelEngine * imEngine = InteractionModelEngine::GetInstance();
 
-    mRunScheduled = false;
-
     // We may be deallocating read handlers as we go.  Track how many we had
     // initially, so we make sure to go through all of them.
     size_t initialAllocated = imEngine->mReadHandlers.Allocated();

src/app/reporting/Engine.h

@@ -69,11 +69,6 @@ class Engine
     void SetMaxAttributesPerChunk(uint32_t aMaxAttributesPerChunk) { mMaxAttributesPerChunk = aMaxAttributesPerChunk; }
 #endif
 
-    /**
-     * Main work-horse function that executes the run-loop.
-     */
-    void Run();
-
     /**
      * Should be invoked when the device receives a Status report, or when the Report data request times out.
      * This allows the engine to do some clean-up.
@@ -131,6 +126,11 @@ class Engine
 #endif
 
 private:
+    /**
+     * Main work-horse function that executes the run-loop.
+     */
+    void Run();
+
     friend class TestReportingEngine;
 
     struct AttributePathParamsWithGeneration : public AttributePathParams