Thank you for additional information. I think I understand your question now.
As I explained in my earlier post, DFT monitors record the time domain electromagnetic field data. Since this data is accumulated over the entire simulation time, it includes all the field data that passes through any specific point. This basically means that for complicated structures that includes reflection from edges or can trap electromagnetic filed inside a geometry, light can pass a point multiple times. Thus, interpreting the phase data for these cases will be quite challenging.
For example, you can try to set a simulation with a plane wave source shined on a substrate at angle. If you fulfil total internal reflection, recorded data at a point will include both incoming and reflected data. Even for such a simple geometry, you will get a very complicated phase data.
Please let me know if you have any supporting reference that verifies your claim i.e. you should not see a phase jump on your profile and I will be glad to discuss it further.