[core] Result node may uses inputs names on creation

[praasz]

Details:

• The Result node may use inputs names when created. Creation such Result assume as previous layer is model output.
• The model when created from outputs and the outputs are note Result node, then create results and take its names as model output names.
• The Result has option to enable/disable use inputs names as it owns. If names not used they still be visible as output names if Result has no dedicated names but if Result is connect to other input these names stay on origin input.

Tickets:

• CVS-159672

[StefaniaHergane]

Tests look good in NPU CI

[dtrawins]

the following models still fail to report the output names:

• tokenizer generated with a command: convert_tokenizer -o bge-m3 BAAI/bge-m3
• dummy model created using ov operations https://github.com/openvinotoolkit/model_server/blob/5b9a3df739789eb71471643a23d2ed1dac4a0ac1/tests/models/no_name_output.py
• yolo11 :

import openvino as ov
det_model = YOLO(f"yolo11n.pt")
ir_model_dir = det_model.export(format="openvino", dynamic=True, half=True)```

The errors from `std::string name = output.get_any_name();` is:

OpenVINO with error:Check '!get_names().empty()' failed at src/core/src/descriptor/tensor.cpp:102:
Attempt to get a name for a Tensor without names.

[praasz]

the following models still fail to report the output names:

• tokenizer generated with a command: convert_tokenizer -o bge-m3 BAAI/bge-m3
• dummy model created using ov operations https://github.com/openvinotoolkit/model_server/blob/5b9a3df739789eb71471643a23d2ed1dac4a0ac1/tests/models/no_name_output.py
• yolo11 :

import openvino as ov
det_model = YOLO(f"yolo11n.pt")
ir_model_dir = det_model.export(format="openvino", dynamic=True, half=True)```

The errors from `std::string name = output.get_any_name();` is:

OpenVINO with error:Check '!get_names().empty()' failed at src/core/src/descriptor/tensor.cpp:102: Attempt to get a name for a Tensor without names.

I can't reproduce issues for: convert_tokenizer -o bge-m3 BAAI/bge-m3 and yolo11.
But can easily reproduce for convert_tokenizer -o bge-m3 BAAI/bge-m3 that without these changes there is no tensor names on model outputs generated by convert_tokenizer

In dummy model example there is no tensor names on model's outputs because names are not set the example has to be modified e.g.:

op0 = ov.opset1.multiply(in1, in0, name="MULTIPLY")
op1 = ov.opset1.add(in1, in0, name="ADD")
op0.output(0).set_names({"mul"})
op1.output(0).set_names({"add"})

or

model = ov.Model([op0, op1], [in0, in1], model_name)
model.output(0).set_names({"mul"})
model.output(1).set_names({"add"})

There is no general feature to auto name tensors in OV, but there are some exception like python API ov.Model where Model creation use Parameters friendly names as tensor names.

[praasz]

@ilya-lavrenov Could you review PR again?