fix: Use proper length for number of total cells in area encoder

[okdistribute]

For a simple area with 3 nodes, I ran the encoder and got a lot of zeros at the end using the latest code on the master branch. (I can send this file to you in a gist if you want).

Looking at the code, it seems like there is a calculation for the size of an area which I don't entirely understand. It seems like it should allocate for the # of cells encoded as varints? I'm not exactly sure what the old code was calculating -- if we need it, probably best we add a bit of documentation or comment about the 🧙 magic 🌟 numbers 👍 😄

EDIT: it isn't a big deal, I think if the length is too long.. then the output could just be trimmed once it's determined that the cell length is actually smaller than predicted.

Before:

03950283e4da06041983c7c26f1623421683c7c2251723421583c7c2f51823421983c7c26f162342010200010e3d4a616d656c277320486f75736500000000000000000000

After:

03950283e4da06041983c7c26f1623421683c7c2251723421583c7c2f51823421983c7c26f162342010200010e3d4a616d656c277320486f75736500