1. You may want to decide early about whether you ship with a cable or not. If you decide on one, you would want to test with it and stay with that. Changing it may require some basic engineering evaluation. If you don't ship, you save the cost and can test with whatever is around, with the potential that a customer may be more likely to encounter a problem.
2. Power supplies are tricky. I've evaluated products with a number of industrial 24V DIN Rail supplies. The testing results they publish are not ever in the same conditions as the actual product, so results with the complete system in the lab can be all over. For instance, it passed with the $100 supply but not the $30 supply.
3. USB supplies are... suspect. Test a few from reputable manufacturers. Ideally something with datasheets that list standards it was tested to. If the USB power supply is not ever shipped, consider rigging up a battery pack to run it directly. Perhaps 4 AA cells in a holder with at most a linear regulator.
4. For radiated emissions, the host computer can be set somewhat to the side. It may be possible to use a longer version of the USB cable to get it more out of view of the antenna. If failures are noted, you may need to test with either the device or the computer off to isolate if peaks are coming from the computer. For backup, bring a computer of a very different model.