[0035] Clarification for threadgroup scope matrix shader stage availability

[haosu-nv]

Which proposal does this relate to?
https://github.com/microsoft/hlsl-specs/blob/main/proposals/0035-linalg-matrix.md

Describe the issue or outstanding question.
In The LinAlg spec, its unclear whether a threadgroup scope matrix is limited to compute shader(and mesh shader), or available to all shader stage?

Additional context
In general, groupshared memory space is tied to thread-group compute model with explicit [numthreads(X, Y, Z)]