I didn't entirely understand. "If I see the bands I will perform four different PCRs to test the best temperature": If you see the bands haven't you already tested the temperature?
I haven't done bisulfite treatment. I thought the goal there was to do sequencing of the product?
You can test your primers in the qPCR machine, or at least with the qPCR reagents, so that the conditions will be the same as the final experiment. Oh, on rereading you are not using a realtime machine, perhaps.
Yes, it sounds good to test the reactions separately first. When combined they may not compete equally. For duplex PCR, which I do, I've needed to adjust the concentrations of the primer sets to get both products to be generated equally. I use a full 20 ul volume, while usually I reduce the volume in single PCRs, because the plateau level of each peak is lower, I think. It will be more difficult with four sided PCR, and you probably want to try the maximum volume reaction. When designing primers, I might order more than one primer set for each target gene so that you can find one for each gene that performs optimally under the given conditions. I was going to add that then you could substitute an alternate primer set if two sets interact to create dimers, but I guess you wouldn't be able to tell which primers were interacting.
Maybe you could first try the individual PCRs, then combine both in tetraplex and also in various duplex combinations to give you some idea of how they compete with each other, and how you will need to adjust the primer combinations.
You're not trying to be quantitative, are you? The duplex PCR I did was for genotyping, so all I needed was a yes/no answer, and the amounts present were only ++ +- or --.