There is an easy or a harder solution for this, depending on whether you know on which vehicle do you want to ad the job to. This solution doesn’t involve constraints.
If you know it:
Simply, create a new problem containing only the single vehicle and the subset of jobs it has in the primary solution and run it again. If you wish, you can feed in the previous route of the vehicle to the problem as an initial route for the vehicle. It will be prety fast compared to the optimalization of the initial solution.
If you don’t know it:
Do the same as in the previous case for each vehicles one by one and pick the one which is the best (the meaning of best depends on your problem space, for example the one with the least cost increase). This looks line a lot of time, but runing the optimalization on one vehicle multiple times is faster than running the original, multi vehicle problem at once. Also it gives you the opportunity to pre filter the possible vehicles (like by utilization level) or to run the per-vehicle optimalization parallel if there are spare cores.