Applying constraints in umbrella sampling

GROMACS version: 2020.1
GROMACS modification: No

Dears,

I have a general question regarding something I keep seeing in MD studies, and still not clear to me.
When performing US, applying a restrain force on one of the pull groups should in principle affect the resulting PMFs calculated by the WHAM, right?
The reason is that I keep coming across works where one of the pull groups is position restrained whereas the second pull group is free to move (example1, example2).

I agree that with such approach one is able to reach convergence faster, but are the resulting PMF values reliable in that case?

Thank you for your responses in advance.

Kind regards,
Ali