I do not know the answer to this, never really looked into how auto-resolve mechanics work. If the engine does a force strength computation based on unit stats, it's possible that the difficulty-based stat adjustments get included in that calculation. Not sure how the tactical adjustments would be included. You might be able to set up a test case with an isolated enemy territory (so the enemy forces can't change). Create an invasion force with 2 possible general options, identical general unit size/type, one with a high command rating, another with low. Save prior to attacking, then attack with the high star general (leaving the low star behind). Check the result, reload, repeat several times to get an average, then do the same with the low star general (leaving the high star behind). If unit strength affects auto-resolve, and the bonuses for general stars are included in that, there should be a noticeable difference in the result.