Auto 3007 Pass image dimensions to detector

[MRo47]

Currently the NN detection output is not to scale with the input image: bug report

It is difficult to pass the input image size to the ImgDetectionConverter as imageManip doesn't have this info seen on this line.

int width;
int height;
if(ph->getParam<bool>("i_disable_resize")) {
    width = ph->getOtherNodeParam<int>(socketName, "i_preview_width");
    height = ph->getOtherNodeParam<int>(socketName, "i_preview_height");
} else {
    width = imageManip->initialConfig.getResizeConfig().width;
    height = imageManip->initialConfig.getResizeConfig().height;
}

detConverter = std::make_unique<dai::ros::ImgDetectionConverter>(tfPrefix, width, height, false, ph->getParam<bool>("i_get_base_device_timestamp"));

This is an ugly fix to set input dimensions values inside the NN.

Maybe a better way could be to set parameters but the API is very limiting on how these can be done?