Since conformations should be able to change during a MD simulation, starting multiple simulations of the same conformation would be a usable tool to conclude if and how a ligand will bind to a protein, also inducing changes to the conformation of both the ligand and the protein potentially.

If you talk about configurations on the other hand, all of them would have to be checked individually as described above.

If you have different binding conformations/configurations later, you can compare them e.g. using the mmgbsa approach or FEP for calculating binding free energies.