Test the HTF with different combinations of nonbonded method

[jthorton]

It would be good to do these changes to the systems before you pass them through the HTF - that way you can test what the HTF does to these forces (and any fixes we might need to do)?

Originally posted by @IAlibay in #1695 (comment)

[jthorton]

We may need to rework the make_htf helper function so that we can apply different nonbonded methods to the input endstate systems and see there effect on the hybrid system.