include [PyTorch FE] quantized::linear_relu

[PT-10]

Details

• Implemented quantized::linear_relu based pimarily off quantized::linear.
• Included pytest

Tickets

• Closes Good First Issue: Support quantized::linear_relu #29733
• Parent Issue: Good First Issue: Operation list to be supported #28584

[PT-10]

@mvafin I've been facing certain issues while running the tests, this corresponds to the following section in ./src/bindings/python/src/openvino/frontend/pytorch/ts_decoder.py :

try:
                # these params exist only for conv params
                stride = pt_value.stride()
                padding = pt_value.padding()
                dilation = pt_value.dilation()
                groups = pt_value.groups()
                res += ivalue_to_constant(stride,
                                          shared_memory=self._shared_memory)
                res += ivalue_to_constant(padding,
                                          shared_memory=self._shared_memory)
                res += ivalue_to_constant(dilation,
                                          shared_memory=self._shared_memory)
                res += ivalue_to_constant(groups,
                                          shared_memory=self._shared_memory)
            except Exception as e:
                logging.debug("Failed to get conv params", exc_info=e)

the tests pass when I comment this out, however they fail when this exists. The issue arises due to the fact that LinearReLU return LinearPackedParams which do not have these params

[PT-10]

@mmikolajcz @PiotrKrzem requesting a review so that we can fix any bugs and move forward. Also having opened the issue in PyTorch, the quantized::<fn_name> seems to be deprecated, and have been moved to pt2e and torchao, let me know if we're to proceed similarly