The antenna is the most critical element in a GNSS solution. If you can accept a position error averaging well above 10 meters with no other obscurations (such as inside a building or a vehicle or with limited sky view) then one of your solutions might be workable. But you still need to assess each.
With SAM, its signal node is in the same plane as the antenna. On its side, SAM will not pick up satellites overhead or in its plane and will have sensitivity peaks in opposite directions perpendicular to its plane. This would be fine if the SAM was stationary, otherwise, with motion and turns, satellites around the horizon will drop as others come into visibility. Worse, ground-reflected signals will be picked up along with direct signals, reducing the effectiveness of direct signals.
A smaller horizontal ground plane for SAM decreases SAM's focus upward and it becomes sensitive in the downward direction and picks up destructive reflected signals. While SAM might work with a smaller ground plane, CAM requires a right-size ground (80x40 mm) to get reasonable signals. A smaller ground plane with CAM could cost you more than 10 dB penalty in signal levels.
CAM is a linear antenna giving it up to 3dB disadvantage compared with circular polarization of SAM antenna. Its only advantage its more omnidirectional pattern (but still with a directional low-signal node)
Given all your design constraints, you will have to try your own experiments under your own use cases to see which gives the best performance against your criteria. But you should also be prepared to relax your physical design constraints to get a reasonably performing product.