Drawson,
One other concern with putting the transducer into the cylinder is that you are actually measuring the position of the piston in the cylinder (where the magnet is) and not necessarily the load that you are trying to position.
We had an instance where the customer would set up the machine in the morning and all was well. It was a spotface application with a tight tolerance on the spotface depth. As the day progressed, the measured spotface depth got larger and larger until it went out of specifications. All the while, the motion controller is reporting that it is dead on position!
We scratched our heads for a while until we realized that the oil was cold in the morning and warmed up with time. The growth in the cylinder rod was enough to push the spotface out of tolerance. The magnet was exactly where it was supposed to be so the motion controller was happy. What was amazing was that when we went to the handbook and looked up the temperature coefficient of steel and did the calculations they were right in line with what we were seeing in the spotface.
We could have added a temperature transducer to the system and done the compensation but decided to change to an external transducer.
If the cylinder rod is short (ours was about 28"), or you control the oil temperature or your tolerances are not that tight then it doesn't matter. But if you have tight tolerances and don't control the oil temperature, you should do the calculation and see if you have an issue with rod expansion.