Crash due to failing assert in SessionHolder::Get

[bzbarsky-apple]

Problem

I just got a crash with this stack:

0x400988de: abort at tools/esp-idf/components/newlib/abort.c:46

0x400e67d9: chip::Optional<chip::ReferenceCountedHandle<chip::Transport::Session> >::Value() const & at connectedhomeip/src/lib/core/Optional.h:170
 (inlined by) chip::SessionHolder::Get() const at connectedhomeip/src/transport/SessionHolder.h:54

0x40105540: chip::Messaging::ExchangeContext::GetSessionHandle() const at connectedhomeip/examples/all-clusters-app/esp32/build/esp-idf/chip/../../../../../../config/esp32/third_party/connectedhomeip/src/messaging/ExchangeContext.h:163
 (inlined by) operator()<chip::Messaging::ReliableMessageMgr::RetransTableEntry> at connectedhomeip/examples/all-clusters-app/esp32/build/esp-idf/chip/../../../../../../config/esp32/third_party/connectedhomeip/src/messaging/ReliableMessageMgr.cpp:147
 (inlined by) Call at connectedhomeip/examples/all-clusters-app/esp32/build/esp-idf/chip/../../../../../../config/esp32/third_party/connectedhomeip/src/lib/support/Pool.h:126

0x401b13d5: chip::internal::StaticAllocatorBitmap::ForEachActiveObjectInner(void*, chip::Loop (*)(void*, void*)) at connectedhomeip/examples/all-clusters-app/esp32/build/esp-idf/chip/../../../../../../config/esp32/third_party/connectedhomeip/src/lib/support/Pool.cpp:97

0x4010516a: chip::Messaging::ReliableMessageMgr::ExecuteActions() at connectedhomeip/examples/all-clusters-app/esp32/build/esp-idf/chip/../../../../../../config/esp32/third_party/connectedhomeip/src/lib/support/Pool.h:257
 (inlined by) chip::Messaging::ReliableMessageMgr::ExecuteActions() at connectedhomeip/examples/all-clusters-app/esp32/build/esp-idf/chip/../../../../../../config/esp32/third_party/connectedhomeip/src/messaging/ReliableMessageMgr.cpp:120

0x401052a9: chip::Messaging::ReliableMessageMgr::Timeout(chip::System::Layer*, void*) at connectedhomeip/examples/all-clusters-app/esp32/build/esp-idf/chip/../../../../../../config/esp32/third_party/connectedhomeip/src/messaging/ReliableMessageMgr.cpp:168 (discriminator 3)

0x40106361: chip::System::TimerData::Callback::Invoke() const at connectedhomeip/examples/all-clusters-app/esp32/build/esp-idf/chip/../../../../../../config/esp32/third_party/connectedhomeip/src/system/SystemTimer.h:61
 (inlined by) chip::System::TimerPool<chip::System::TimerList::Node>::Invoke(chip::System::TimerList::Node*) at connectedhomeip/examples/all-clusters-app/esp32/build/esp-idf/chip/../../../../../../config/esp32/third_party/connectedhomeip/src/system/SystemTimer.h:224

0x401063c4: chip::System::LayerImplLwIP::HandlePlatformTimer() at connectedhomeip/examples/all-clusters-app/esp32/build/esp-idf/chip/../../../../../../config/esp32/third_party/connectedhomeip/src/system/SystemLayerImplLwIP.cpp:139

0x40109202: chip::DeviceLayer::Internal::GenericPlatformManagerImpl_FreeRTOS<chip::DeviceLayer::PlatformManagerImpl>::_RunEventLoop() at connectedhomeip/examples/all-clusters-app/esp32/build/esp-idf/chip/../../../../../../config/esp32/third_party/connectedhomeip/src/include/platform/internal/GenericPlatformManagerImpl_FreeRTOS.cpp:157

this is an m5stack on sha 5053f20. I was doing a toggle command using chip-tool. Clearly we have an exchange context without a session handle at that point...

Proposed Solution

Figure our why we can be in this state and fix things to not crash.

@yufengwangca @kghost