There's a lot of tracker tradition in pattern length 64, but musically speaking, it favors less common note subdivisions: If we imagine the 4-beat whole note as the "most common" subdivision, and more complex ones appearing as their ratios require larger prime values in their denominator(1/2 is more complex than 1/1, 1/3 is more complex than 1/2, but 1/5 is more complex than 1/6) then it would make sense to have the default aim primarily for handling divisions of 2 and 3 comfortably. But with a length of 64, we leave divisions of 3 out in the cold.

Thus it works out that I frequently realize while composing that I want to throw in a triplet, but that I'm already locked in on the 64/16/4 default, so I end up altering the rhythm rather than try to rewrite the pattern. Although 16ths are also fairly common, "Grow Selection" already favors them since it doubles the row spacing, so it'll fix any multiple of 2. So it remains true throughout that a better workflow with less fuss would result from starting at 96. (Another possibility is to have an configuration for "Grow Selection" that triples row spacing, but that makes the UI a little more complicated in acting to ease the much less common event of desiring a song with 1/9th or 1/27th divisions.)

I'm not just theorycrafting on this, too: Most MIDI implementations will use a PPQN value of 96 as well, so someone at some point thought this through with respect to quarter note divisions, and reached a similar conclusion.