this seems antithetical to the idea of Langton's Ant
well, the optimal strategy is clearly to place the food close to the starting block, which ends the simulation fairly quickly. As it is now, this requires basically zero understanding of the movement code.
the key idea of Langton's Ant is about long term behavior, so this also seems antithetical to the idea of Langton's Ant.
since you are able to edit the ant's code in the first few levels it seems like it would make more sense for the food to be pre-placed somewhere far away, and for you to need to edit the code to make the ant go there.