The common practice stems from the widespread-but-fallacious belief that loaded spinal flexion inherently causes injury, and this belief originates largely from studies of intradiscal pressure in standing, seated, or otherwise spinally-flexed positions of untrained individuals. It is often assumed that we should avoid any and all flexion, extension, or rotation whilst loading our spines—that we should lift very straight and mechanically. And this is true when we are speaking about maximal and near-maximal lifting. However, this simplistic view of the spine fails to account for the fact that it is designed to articulate under load, and that it is quite capable of doing so healthfully within a range of loads.
Virtually all sports and activities involve some degree of simultaneous flexion/extension and rotation under load, and those actions are critical to their performance. Running, jumping, ducking, weaving, changing directions, tumbling, grappling, and throwing—all of these would be impossible without multi-planar movement of the spine under load. And since performance sports all demand these actions, it follows that they can and should be trained.
The caveat is the degree to which the spine can safely be flexed, extended, and/or rotated under load, and of course, the amount that it can thereby be loaded. The greater the angle of flexion, extension, or rotation of the spine, the greater the stresses—axial, shear, and rotational—that compound on the intervertebral discs. High degrees of flexion or extension, particularly, cause high intradiscal pressures (hydrostatic pressure measured in the nucleus pulposus) to act against the disc wall that is under tension. Hence, the closer that our training load approaches our maximum for any given lift, the less we should allow our spine to deviate from neutral. However, the deep spinal muscles of the interspinales, intertransversarii, rotatores, and multifidus combine with the superficial erector spinae and ‘core’ muscles to counteract these stresses. Thus, we can safely assume that if we train these structures sensibly, it will reduce the likelihood of future injury.
Thus, loaded appropriately—that is, not loaded excessively—the Jefferson curl can be an excellent exercise for stretching and strengthening the posterior spine. Since it involves a large degree of spinal flexion, however, the load used should be low—perhaps a maximum of a quarter or third of our deadlift one-repetition maximum. And it is implicit, therefore, that this exercise is more appropriate for training the deep spinal extensors than the large superficial extensors of the erector spinae.
I hope that helps.