Robotics: What are the advantages of cartesian paths in MoveIt?

Question

I have some experience in C++ programming but I am a bit new to robotics. I have to create robot functionality for a customer, where the robot arm moves in between several objects to reach a given pose goal for its end-effector (gripper).

I have to use the MoveIt interface for motion planning. This interface can also compute Cartesian paths. If I understand correctly, cartesian paths are just a colletion of waypoint poses that the end-effector should reach sequentially.

My first question is, what is the advantage/disadvantage of executing a cartesian path, rather than moving to several waypoints manually one by one?

My second question is, would a motion planner first plan all the waypoints and then start executing them them one by one. Or will a motion planner repeatedly plan and execute for each waypoint in the cartesian path?

The reason that I would like to know this is because I think that moving the end-effector in steps (or waypoints) would improve the chances of finding a trajectory for a goal that is very difficult to reach due to many objects being in the environment.

Thank you in advance,

Dirk