openvino: [Bug] Memory Leak on inferance

System information (version)

• OpenVINO = 2022.1.0.643
• Operating System / Platform => Windows 64 Bit
• Compiler => Visual Studio 2019
• Problem classification: Model Conversion
• Framework: pytorch(if applicable)
• Model name: OCR (if applicable)

Detailed description

when i use OpenVINO2022.1.0.643 to inferance a text image “infer_request.infer();” always Memory leaks. code list: ############################################################################

######################################################################

Steps to reproduce

```.cpp
// C++ code example
```
reco_network400 = core400.read_model(string(modelPath), modelPathBin);
auto d400 = reco_network400->input(0).get_partial_shape();
d400[3] = -1;
reco_network400->reshape(d400);
const ov::Layout tensor_layout400{ "NHWC" };
ov::preprocess::PrePostProcessor ppp400(reco_network400);
ov::preprocess::InputInfo& input_info400 = ppp400.input();
input_info400.tensor().set_element_type(ov::element::f32).set_layout(tensor_layout400);
input_info400.model().set_layout("NCHW");
ppp400.output().tensor().set_element_type(ov::element::f32);
reco_network400 = ppp400.build();
compiled_model400 = core400.compile_model(reco_network400, "CPU");
infer_request400 = compiled_model400.create_infer_request();
......
    image.convertTo(image, CV_32FC1, 1.0f / 255.0f);
    image = (image - 0.588) / 0.193;
    vector<vector<float>> output;
    size_t size = long(width) * long(height) * long(image.channels());
    std::shared_ptr<float> _data;
    _data.reset(new  float[size], std::default_delete<float[]>());
    cv::Mat resized(cv::Size(width, height), image.type(), _data.get());
    cv::resize(image, resized, cv::Size(width, height));
    ov::element::Type input_type = ov::element::f32;
    ov::Shape input_shape = { 1, (long unsigned int)height, (long unsigned int)width, 1 };
    //ov::Shape input_shape = { 1, 1,(long unsigned int)height, (long unsigned int)width };
    ov::Tensor input_tensor = ov::Tensor(input_type, input_shape, _data.get());
    infer_request400.set_input_tensor(input_tensor);
    infer_request400.infer();
    ov::Tensor& output_tensor = infer_request400.get_output_tensor();

Issue submission checklist

• I report the issue, it’s not a question
• I checked the problem with documentation, FAQ, open issues, Stack Overflow, etc and have not found solution
• There is reproducer code and related data files: images, videos, models, etc.