I guess you are concerned about relaxation and MD simulations.
We use a mixed strategy. For the forces we do not impose symmetry constraints, however, if you start from a symmetric configuration,
usually the calculated forces will obey this symmetry to a high accuracy and you will stay in this symmetry.
For the stress we perform a symmetrization, thus for cell relaxations the symmetry will be fixed.
Thus, we keep symmetry in relaxation, but for finite temperature md we do not keep the symmetry. There is no option to explicit impose symmetry constrains for the coordinates.
To remove the symmetry constrain, add a small random number to each coordinate and cell vector. In the builder, the tool bar icon of a dice does this for the coordinates,
for the cell vectors you need to add the number you self.