[BBray_T1296]

Churro tubes are a Solid Mechanics example of one of the worst cross-sectional profiles imaginable to transmit torque.

Internal shear stress distribution in a pure torsion model is linear with radius. That is to say the further-out parts of the shaft (from the neutral axis) carry more rotational shear stress than the further-in parts.

In a circular shaft, from the center, the 0 to r/2 part of the shaft handles (1/2)^4 of the torsion (6.25%) while the outer half (r/2 to r) of the radius carries the remaining 93.75%.
This means a shaft with a hole in the center of half the diameter is only 93.25% as stiff, but with %75 the weight, making it a reasonable trade-off.
This also means a solid shaft with half the radius will deflect the same amount (radians/unit length) as the full shaft with just 6.25% the load and 25% the weight.

The churro tube with its "snowflake" cross section gains its weight savings by carving out the outer sections of the shaft. This is removing much of the primary torsion carrying material, leading to a horrendous torsional rigidity. I don't have the time to run numbers unfortunately, but it has to be really pitiful.

Note that in tension/compression or beam shear proximity to the neutral axis is irrelevant, making a churro tube no different than any other shaft with identical x-sectional area in these properties.
Beam bending moments are probably pretty decent due to the I-beam-like sections so churro tube may be a reasonable choice there

tl;dr Churro tube has an exceptionally poor torsional stiffness per unit weight, for weight savings use a hex axle with a hole drilled in it instead