I see your concerns. Here are some recommendations which may overlap with what you already have read, but for completeness sake, I will include them here.
1. Include the stress and strain data on the rst. This is pretty straightforward.
2. Mechanical and ACP have stresses reported differently. ACP reports in the fiber/matrix and results are a single value per element. Mechanical can report in the Global or local or Fiber/matrix and you will get the distribution based on the gauss point result variation, hence not a single value. Be careful in how you are comparing. I recommend a simple short beam shear specimen (like 3 point bending) and you can compare with theoretical hand You should see very close agreement between hand calc, shell and solid in ACP and Mechanical, provided the stress gradient is not an issue and you have the proper mesh resolution. This is an exercise I would not skip so you know you are looking at the right quantities in the right way as your baseline.
3. If there are large variations in the material properties, the most accurate ISS will come from modeling each ply as a layer with solid elements. For sandwich construction I would recommend at least a different element for the facesheets versus the core and if the core is thick, I would tend towards solids.
4. The ability to compute INS for shells is a nice feature and won't match exactly but depending on the size of your model as you may not want to model with solids for computational reasons.
5. The recompute iSS is a nice feature to smooth out the ISS and looks closer to theoretical results, but it does assume the shear force on the surface is zero so if your composite body does not have free surfaces, use caution and don't use it.
I hope this helps. Thank you!