Yes! This is true, depending on the driver in question. Sorry, I should have clarified but I didn't really want to get into it on that thread.
I think most people know that a headphones rated impedance is a nominal value, meaning the impedance is not constant across the frequency range, how flat the impedance curve is is dependent on the specific driver. With dynamic drivers, there can be resonances at low and high frequencies where the impedance will spike. Something called a voltage divider is formed between the output impedance of the amplifier and the impedance of the driver - essentially the output voltage of the amplifier is split across these two impedances, how much voltage is dropped across each will depend on the ratio of these impedances.
So, if the driver is being run by an amplifier with a relatively high output impedance, you may see a bump in the frequency response at these resonant frequencies as more voltage is being dropped across the driver due to the impedance spike at that frequency. If the output impedance of the amplifier is very low, then the impedance spike at these resonances becomes negligible in terms of frequency response since nearly all of the voltage will be dropped across the driver.
That's a pretty basic explanation without using math, but I hope it makes sense. There is tons and tons of info out there just on this subject if someone wanted to dive deeper into it.
So yes, that is one way tube rolling can affect the frequency response IF you are changing output tubes to different types that will significantly alter the output impedance of the amplifier. However, I don't think that explains a lot of the differences people hear when tube rolling, especially when swapping input tubes.
Often you will see people say "X tube is warm and Y tube is bright" even if the two tubes are the same model, say two different 12AU7 for example. If you were to measure the frequency response of the amplifier with these two tubes, you would most likely find it measures exactly the same with both! So why is there a perceived difference?
This is bit of a mystery, and I'm sorry but I don't have a definitive answer, but my suspicion is is related to the distortion characteristics of the tubes having a psychoacoustic effect. Since tubes are imperfect devices, if measured you will often see differences in THD between two tubes of the same type, even if they are both NOS measuring, which certainly is audible and will affect how someone perceives the sound.
Also, there may be some contribution from material and construction differences of the tubes. I think you'll find this isn't something that has been studied in any scientific way, but it seems to me there is enough anecdotal evidence to suggest plate / grid / cathode material and construction does contribute to the character of a tube's sound in a meaningful way.
But the point I was trying to make is that these differences truly are perceived, the amplifier's frequency response is not changing in any measurable way when you swap tubes of the same type!