For "Partial-Profile" designs, our software does not formally support the idea of have partial-profile & alternative-specific designs (at the same time). Some people have been able to "fool" our software into doing this, but please be careful as you may get unexpected and wrong experimental designs. You would need to check each design version carefully to make sure that it produced what you were expecting in terms of valid product combinations. And, you should check the "test design" report to make sure all the standard errors for simulated (robotic) respondents were reasonably small to imply good precision (0.10 or less for alternative-specific attributes; and 0.05 or less for generic attributes).
As you indicate, creating standard prohibitions to impose alternative-specific designs is typically much less efficient than doing the alternative-specific design approach. Often, using standard prohibitions to impose alternative-specific designs leads to so much multicolinearity in the experimental design as to make the data unusable.