1. Skyline always subtracts the background area from the total areas. The background level of a peak is assumed to be the intensity at the start of the peak or the end of the peak whichever is lower. There are some diagrams showing how background is calculated here:
https://skyline.ms/wiki/home/software/Skyline/page.view?name=tip_peak_calc
2. No, there is no way to do what you are asking. Note that when you choose "View >Transform > Savitzky–Golay Smoothing" that only affects the way that the chromatogram is drawn on the screen. The peak areas are always calculated based on unsmoothed values.
3. No, Skyline never tries to tease apart peaks that overlap in the time dimension.
4. No, Skyline does not use retention time alignment to automatically pick peaks, but I am supposed to make this better in the next version of Skyline. Here are some things to look forward to:
a) You can already use the menu item "Edit > Integration > Apply Peak to All" to choose peaks in the other replicates based on what has been chosen in the current replicate. Some retention time alignment happens while Skyline is figuring out which peak in the other replicates corresponds to the current replicate. Currently, "Apply Peak to All" only works on one peptide or molecule at a time, but, soon you will be able to select a lot of things in the Targets tree and have the algorithm work on all of them.
b) We are also supposed to make alignment and peak picking work together in other ways, but we have not yet figured out how exactly that will work.