It could be useful to have modulation synced to a sample or a part of a sample. You might compensate for differences in volume or pitch at the beginning or ending of a loop. You might have interesting effects operating in sync with ping-pong looping. You might have modulation that behaves as though it were a part of the sample, progressing faster with higher pitches and slower with lower pitches.
To put it another way, I’m describing a modulation envelope where the x-axis is not beats or milliseconds but rather samples.
Ideally there would be something akin to the ADHSR device to make setting values for the entrance and exit of a loop easier.