Good points I'll grant, but it naively seems to me that this issue with batteries running low is a perennial problem with spacecraft.

Unless the RTG/solar mass ration is very high, your points about having most of the experiments already ran in the first day, or about RTGs not helping with other failure modes, don't matter too much. Either way when you include a power source (RTG or solar) you are doing so for the purpose of getting data after the first day.

Besides avoiding the shading problem, I believe RTGs are also more robust to unfurling issues and to impact damage.