AS I understand it SCiO can only work on the reflected NIR of the molecules that are illuminated. So estimation of elements in a complex and varying mix would not work. I cannot imagine how you would set up the reference samples.
Pesticides are likely to be well below the 1% detection threshold. This has been a strong wish from many, but CP have repeatedly said that it is unlikely to work. I do wonder if the samples could be worked up to concentrate the surface chemicals – washed in alcohol for instance, and then the alcohol evaporated to give a “strong” solution. If accurate contamination levels were required, getting known reference samples would be tricky without access to good lab facilities. It is such an important topic that it should be investigated, even if the success likelihood looks low.