Question: How can the proposed LUVOIR space telescope slew to different directions while keeping the sunshade in a fixed orientation? What compensates?
Also:
I mention that it moves as a rigid body; in order to change the direction the telescope is pointing the whole spacecraft slews, including the sunshade.
In layman terms i think you are asking what is the big difference between JWST and LUVOIR, in terms of how each slews to observe its target.
As you say, JWST moves as a rigid body, ie. as one. LUVOIR does not.
LUVOIR is composed of two elements - the spacecraft and the payload:
payload is the telescope and the part that needs to be vibration free and it is this that slews around.
spacecraft is the support structure that houses avionics, fuel, sun shield, and the 4 CMG's, and everything else that generates vibration as well as providing attitude control of the entire system. The payload element determines the spacecrafts attitude.

The key element of this concept is VIPPS, which allows it to have a disturbance free payload (telescope):
The highest possible degree of isolation of a sensitive payload structure from spacecraft disturbances is realized through no physical contact between the two bodies.
Lockheed Martin Space Disturbance Free Payload technology, in development since 1999, has resulted in the basis for the non-contact Vibration Isolation and Precision Pointing System (VIPPS) for LUVOIR.
VIPPS uses voice coil actuators, which do not contain any moving mechanical parts, where an axial force is generated between a permanent-magnet field assembly (mounted on the telescope payload side of the VIPPS interface) and a coil-wound bobbin (mounted on the spacecraft side of the VIPPS interface).
Non-contact sensors at the VIPPS interface provide a real-time
measurement of the interface relative translation and rotation; this measurement is used in the VIPPS control system to maintain stroke and gap at the interface.
Technology readiness level 6, demonstrating this technology, is proposed for a CubeSat launched sometime before 2025.
Disturbance Free Payload allows payload and spacecraft to fly in close proximity without physical contact, using custom-designed, large-gap non-contact actuators.
A DFP-configured system is actually two spacecraft flying in close formation.
This was patented in 2002.
To control its attitude, the telescope pushes against the supporting spacecraft using a set of six noncontact linear-motion, electro-mechanical Lorentz force actuators.
Attitude of the telescope is determined using a Fine Guidance Sensor or other LOS sensor on the payload, and the error signal garnered from six non-contact position sensors is used to drive reaction wheels and thrusters on the supporting spacecraft.
The Vibration Isolation and Precision Pointing System (VIPPS) enables the telescope to achieve extreme pointing and image stability while still meeting the line-of-sight agility requirements consistent with its astronomical Surveyor goals.
The spacecraft controls its inertial attitude such that interface stroke and gap are maintained. Since the telescope is physically separated, the disturbances and structural excitation of the spacecraft and sunshield do not propagate to the telescope.



In these examples, the lateral position of the telescope center-of-gravity remains constant. This constraint requires both ends of the boom be articulated as illustrated above. The vertical distance between the observatory and spacecraft centers-of-gravity do change because the VIIPS boom is of fixed length.
For large-angle slewing, the 4 CMGs located on the supporting spacecraft body are used.
The mechanical interface between the telescope and the supporting spacecraft houses 6 vcm and associated sensors located between the gimbal and backplane support frame.
The payload section itself has vcms and associated sensors for the telescope itself.
http://surveygizmoresponseuploads.s3.amazonaws.com/fileuploads/623127/5043187/119-806ddbfd3b26fd17af7803ce18b5cf5f_NordtAlisonA.pdf
http://www.mrbolcar.com/uploads/1/0/6/7/106798055/103980b.pdf
https://www.hou.usra.edu/meetings/landscape2019/presentations/Nordt.pdf

The gimbal and boom can be seen on the back here:

https://www.spiedigitallibrary.org/conference-proceedings-of-spie/11115/2528190/Dynamic-wavefront-error-and-line-of-sight-performance-predictions-for/10.1117/12.2528190.short