Unfortunately, I don't see how recording a starting state could work. The input can be anything, from MIDI, music, wave shape, etc. In other words, you can't always compute what the value will be in n seconds since an input like the music isn't predictable.
Maybe an option to avoid resetting the grid each time it is deactivated? This way, when the grid is activated again, it will continue from where it left. However, you wouldn't able to always start with the same values. So I'm not even sure it's a good option.
If you're using sin, cos or another wave shape, why not use the phase shift?
Alternatively, to avoid deactivating the grid, you can set its activation to a very low value like 0.01%. It's easier than using multiple precedence levels.