I think I understand what you are saying about the database, but I will just explain what is held to avoid possible confusion, including anyone who may be following this thread. Where the observer provides the instrumental magnitudes and uncertainties/errors of the variable and comparison stars then this data is recorded in the database. Thus, if someone wants to recalculate the magnitude using different reference magnitudes, or to exclude one of more of the comparison stars then this is possible. It does not hold information on non-comparison stars, e.g. every star in an image, nor does it hold data on background counts etc. We only require the Julian Date to be submitted, noting HJD and BJD can be calculated, and as a general rule data that can be calculated from existing data is not stored. The exception to this is the derived magnitude, as different observers could calculate this by different methods, and it is desirable to hold what the observer calculated.
As you are doing, it is a good idea to store images including the calibration frames. Then if a recalculation is required, or there is a query from a researcher, it is possible to delve into the detail.
There is no easy or right/wrong answer on whether to submit individual images or combined images. Individual images allow researchers to choose how to combine data (within the limitation of not having the raw images to hand), while the observer doing this can make life easier for the researcher. Also, the observer will be most familiar with their setup and the images. Personally I would suggest combing images if your results are dominated by the noise.