Hello Luis,
Perhaps I explained that poorly as convergence/divergence is only remotely related to what I was trying to get at. You are injecting a rectangular field larger than the domain so are expecting a uniform distribution across the input. The important part is to look at your observation area and note that the field and therefore the Poynting vector (what we integrate to get power) is not uniform. Because it is not uniform I would not expect something that is integrated across it (the power from the observation area) to match a measure based on one mesh point at the center of the field (observation point).
Another illustration of what I am trying to get at is to double SIZE parameter you have set doubling the transverse direction and rerun your simulation and extract your values. You should notice that the results from the observation area are effectively the same while there is significant changes due to the observation point. You are correct that the observation area is what you should be using.
As to why it is not showing the expected reflectivity I am not certain, do you have a paper you are trying to reproduce? One thing I did not is can you please confirm what materials you are using because the refractive indices seem off to me but could be my misunderstanding your material naming convention.
Please be aware the 32-bit product is an outdated release that is no longer receiving software updates and as such can only receive limited support at this time.