Very interesting information you guys found! If the impedance is the same, the current will more or less be close enough to where it won't matter! I am guessing the biggest reason the 70W ballast is physically beefier than the 58 ones is that they need to be able to dissipate a little more heat. I will definetely try some 58 watt tubes on these ballasts, once my install is done

I will definetely do some measurements, because now I am curious. a 5% overdrive should be nothing to worry about, that can happen in normal fixtures too if the mains voltage is high. It should not affect tube life by much.
Input voltage matters a lot too, of course! I am very sure that at 220V, it will run a 58W tube all day without even the slightest hint of overdriving. BEcause I have tested one of my 70W tubes already and at 220V, it underdrives the tube according to my wattmeter on the input. It draws around 63 watts at 220V, so that is probably around 58W to the tube

at 230V it draws slightly more, but still probably not to the point of where it would overdrive a 58. Still definetely underdriving my 70 watters at 230! I will adjust my inverter voltage (I have a fully adjustable victron inverter powering my shed, its all off grid and solar powered!) to around 240V once I have the two 70 watters wired up in the ceiling

I definetely do not wish to underdrive them since that is probably bad for them, much more so than overdriving slightly would be! Underdriving causes electrode sputtering which is not good!
This is definetely interesting, because this means that I should be able to swap my 70 watt tubes for 58 watters and run those just fine too without overdriving them! And this is something I can actually do, since I am using C-clips to hold the tubes , and the sockets are free floating ones that hold themselves onto the ends of the tubes. The ballasts are designed for sun tanning tubes I believe, but I am doing something a bit different with them

I have a decent idea how to make this install quite neat, and it will definetely be cool.