For that last example, "Practicality and Cost for a Supermarket Petrol Filling Station to be Replaced by Grid Hydrogen Production"... the caption correctly notes that upstream emissions should be factored in. That's also the case for a pure EV.
But... if you were to add a row or two to your table examining Grid CO2 emissions per mile for pure EV, I think you're going to find lower numbers for the pure EV. Heck, throw in a row for Oregon or Norway or somewhere where the grid is really clean, to show what we're aiming for.
Not to mention the "minor" point that the filling station requires a 13 MEGAwatt connection, while the pure EVs are being charged at home, at workplaces, and perhaps occasionally at "filling stations" (e.g., superchargers).
Moreover, as the grid becomes cleaner, each EV **already deployed** becomes much cleaner. Ironically, with on-site steam methane reforming (SMR) (perhaps the cheapest case for a hydrogen vehicle), as the grid becomes cleaner, we'll still be stuck with CO2-dirty SMR.
Thank you for all the work you're doing assembling this analysis!
Alan