fix(mge/tensor): fix mem leak when getting value in trace

GitOrigin-RevId: 8bca2e738016f9ed45a3c413b1024976ea0cd952

imperative/python/megengine/jit/tracing.py

@@ -1002,8 +1002,7 @@ class CompiledTensorProxy:
             else:
                 # c++ will throw TraceReadError
                 return None
-            if self._isscalar:
-                self.__value = self.__value.squeeze()
+        # c++ side will handle scalar case
         return self.__value
 
     def _dev_tensor(self):

imperative/python/src/tensor.cpp

@@ -393,7 +393,9 @@ PyObject* TensorWrapper::numpy() {
             throw TraceReadError("value of this tensor is not read in trace");
         }
         if (m_tensor->m_flags & Tensor::Flags::SCALAR) {
-            np_val = PyArray_Squeeze(reinterpret_cast<PyArrayObject*>(np_val));
+            PyObject *np_scalar = PyArray_Squeeze(reinterpret_cast<PyArrayObject*>(np_val));
+            Py_DECREF(np_val);
+            return np_scalar;
         }
         return np_val;
     }