Not the issue exactly.
I've done a cursory run through Bullet's own documentation, and reviewed some of Godot's module source for it. The configuration of bullet's joint differs when node_b() is absent or not. When present, node-b's transform is massaged along with node-a's and passed to the bullet constraint constructor.
Somewhere there was mention that if node-b is not present that the constraint 'operates as if there was a static body in place' or something to that effect and it is the case as a normal (dynamic) rigid body as node a will free in place with a default 6dof, this absolutely works but i had thought that manipulating the joint's global transform would effect the constraint's orientation, as if there was a kinematic rigid body paired with it was being manipulated.
but nope, the constraint's transform is set in stone when only node_a() is present
The high order bit was I was hoping not to have to have a kinematic specifically to manipulate the placement and orientation of the constraint reference frame itself.
--edit--
As a note for anyone who might be trying to figure out where the pivot of a joint is defined, the relations between the rigid-body/ies and the pivot joint are defined when node_a/node_b are set on the joint itself.
There after the only way to manipulate the relative orientation of the bodies or a body is by manipulating the constraint parameters (linear and angular axis properties) or manipulating one of the bodies (via forces if you want to keep the physics-solver happiest) or (more or less and again physics-solver friendly) whatever you want by manipulating a rigid-body set to kinematic (with in reason)
