Friction in PhysicsMaterial is too low even at 1.0 (the max)

[DanielKinsman]

Godot version:
3.3 stable

OS/device including version:
linux

Issue description:
At the maximum value for friction (1.0) things are still very slippy. It was clamped to 0-1 in #42202. Experimenting by forcing a value of > 1.0 by manually editing the tscn with friction = 100.0 provides the grippyness I am after. Is it possible that the docs are wrong and it is actually a percentage rather than normalised? Or am I just abusing undefined behaviour in the physics engine? Values >100 seem to behave the same as 100 but I haven't done extensive testing.

Steps to reproduce:
Run the minimal reproduction project. One of the objects has a friction of 1.0, the other has a friction of 50.0 (both set with roughness). The one with 50.0 friction behaves with much more grip.

Minimal reproduction project:

frictiontest.zip

[akien-mga]

CC @rcorre @godotengine/physics

[madmiraal]

Despite what the documentation suggested, there is a case for both bounce and friction to be greater than 1; so I think
#42203 and it's cherry-pick, should be reverted.

Note: Although, as pointed out here, combined bounce is clamped by the physics server to be between 0 and 1, the same is not true for friction:

godot/servers/physics_2d/body_pair_2d_sw.cpp

Lines 214 to 220 in d578d82

	real_t combine_bounce(Body2DSW *A, Body2DSW *B) {
	return CLAMP(A->get_bounce() + B->get_bounce(), 0, 1);
	}
	real_t combine_friction(Body2DSW *A, Body2DSW *B) {
	return ABS(MIN(A->get_friction(), B->get_friction()));
	}

[rcorre]

I wonder if it is different between 2D and 3D? I tried putting cubes on sloped planes with friction values of 50, 1, and 0.1 (both the plane and the corresponding cube). 0.1 slides down the slope, 1 sticks to the slope, and 50 jitters erratically.
example.zip