I disagree with all you have said. Sorry.
I spent many years in failure analysis so I understand the life cycle of wet electrolytic caps pretty well. It seems the bigger the can, the more susceptible they are to drying out. Many very old amps the caps are dry paper and do not age. The industry standard for large electrolytic is about 7 to 10 years. Heat is the critical factor. The higher the ripple current, the warmer. Where stored can make a difference. Six months on the top of a high-bay in Colorado, they are toast. FWIW, the 4200's in my 20 year old preamp were not to spec, but the 5200's in my little integrated are just fine. If an amp is never used, and sits in AC. it will last far longer.
On overkill. Absolutely not. 4 x 6800, not 68,000. So 27,000uF total per rail. I am not talking about ripple, but in maintaining the rails during transients. This makes very big audible differences. It seems Rotel shares your view and this may explain why they are totally lifeless to listen to even though their architecture is very good.
I offer this possibility: The bank size not related to the amp power, but to the dynamic current needed. If I only need 60W without clipping, it does not matter if the amp can produce 300, the dynamic current is the same. So, how big a rail supply is has more to do with the power required to not droop the rails, not power available. I have a suspicion this is the primary reason so many find big amps better sounding than small. It is the dynamic current. If you look at the best of the smaller amps, you will find proportionally larger banks. ( There are a few other things too) Also remember, for the same uF, if you double the voltage, you double the coulombs stored.
Besides, in my amp, the ripple to the output side is more than 110 dB down. I use a pi-filter to the VAS and IPS so I get another 10 dB there. Good enough.