pull down to refresh

I agree that fiat-denominated fees have are not as low as btc-denominated fees. See the second chart in the OP.
When it comes to miners who have fiat-denominated expenses, this argument may make sense (...)
(...) affected by the change in exchange rates, and so it seems to me that they are moot.
That's why I explicitly mentioned 'value' and differentiated it from 'price'. Your points seem to focus on 'price'. My whole reasoning focuses on 'value'.
A metaphor could be Mass vs Weight. Mass is inherent property of a body. Weight is a measurement you get using something else. And on different planet you will get different weights, while mass stays the same. Similarly, if you use different denominating assets (like dollar, euro, eggs or 1 kg of beef) you will also get different prices.
Value is like mass - it's an inherent property of an asset. Price is like a weight - it's a relative measurement when you express value of an asset denominated in a value of other asset (UsdBitcoinPrice == ValueOfBitcoinUnit / ValueOfDollarUnit).
My claim is that not only 'price' of bitcoin increased, but also 'value' of bitcoin increased. And it's obvious why is that the case - adoption is bigger, there are more use-cases, the track record is longer - those increase the inherent value of bitcoin unit.
Hope that clear things up. Todays sats have more value than 2011 sats, so we need to pay less of them for the same thing.