In theory changing the bad cell out for a new one works fine. In practice, you end up with a battery pack that is only as good as one of the other weaker cells. The reason one of the cells in a pack fails, is often because the voltage was allowed to go below the minimum safe level. If the voltage goes low enough, all sorts of weird chemical things happen. By replacing that bad cell, you get the pack back, but the rest of the cells are most likely compromised as well. Since the pack is a series combination of cells, if any one of the other cells has a higher internal resistance as a result of a charge or discharge issue, then the whole pack will have a lower overall capacity. The pack won't take a charge properly and it will run out of charge even though it appears to have lots of voltage capability.
You can't change just one cell in a pack for the same reason you don't mix brands or sizes of batteries. They have slightly different parameters and they don't play well together.
If you want to reclaim a bad cell, put a controlled high voltage charge on the bad cell. I have used a 2V charge on the 1.2V NiMh cells with success. I monitor the heat and current input to the cell. As long as it is taking the charge and not getting too hot, I let it go. As soon as I detect an increase in the heat, I stop. Quite often, I can salvage a bad cell like this. It will go from dead back to useable and can be replaced back into the pack it came from.
Your idea is still valid. I do like having a pack that uses regular NiMh cells. I was contemplating making one myself. You just have to operate it like one large pack and not as individual cells, although you can charge the cells individually if you want. Just make sure they are all topped up fully or balanced each time they are used.
Except for the Automag in front, its usually the man behind the equipment that counts.