Locked constraints with multi-zone problem and planning units with NA values in some but not all zones

[jeffreyhanson]

The add_locked_in_constraints and add_locked_out_constraints will fail--and ungracefully--when locking planning units in a multi-zone problem where some cells have NA values in a given zone and non-NA values in a different zone. The current fix is to simply set "real" planning units to have finite values for cost and manually lock in and lock out planning units using the locked in and locked out constraints. This is not ideal, and users should be able to simply supply locked in constraints and planning unit data where some planning units have NAs in some zones and non-NAs in other cells. Additionally, the documentation for add_locked_in_constraints and add_locked_out_constraints is not very clear about how the locked data should be specified in such cases.