It has come up in the past, where the volume that had been automatically set in training was not the desired volume, so having VoiceAttack set it after the speech engine starts would likely be useful(More flexible than adjusting it using the training process at least).
I have advised people to use a command that sets microphone volume when VoiceAttack starts to achieve the same effect, but having it built-in would make it profile-independent, and generally more convenient.