What follows is not specific to direct positive paper, but in general applies to preflashing of silver halide materials; please keep this in mind. However, the same principles will apply to direct positive just the same.
Silver halide grains need a certain amount of exposure to become developable. You can think of this as a threshold that the silver grain needs to be pushed over before it can create density during development. Beyond this threshold, there is a more or less proportional relation between exposure and density buildup. Up to the threshold, there's no relation between exposure and density; any exposure that remains below the threshold will not result in density.
Keep in mind furthermore that exposure is cumulative; the silver halides 'remember' (to a large extent) how much exposure they have received. This allows us to compound a total exposure through consecutive, shorter exposures.
Flashing is non-image-wise exposure that helps the silver grains some way (or even entirely) towards/over this exposure threshold. If you combine all of the above, you'll realize that the influence of flashing on image density is highly non-linear. Flashing has a proportionally very large effect in areas that receive little image-wise exposure: after all, in these areas, flashing can make the difference between no density at all, and some visible/measurable density (effectively an 'infinite' increase). However, in areas where there's a lot of image-wise exposure, the added density given by flashing is marginal and ends up being invisible. As a result, flashing essentially only affects the lower part of the density curve (so the brightest highlights in a positive-working material, or the deepest shadows in a negative-working one).
So from a practical viewpoint, assuming a positive working material, what you can achieve by flashing is compression of the highlights. It will not affect the shadows. Taking this back to your statement in bold concerning shortening exposure: your statement is correct only insofar as it relates to highlight detail. In terms of the rest of the image, flashing has virtually no influence. However, this can be a very desirable situation in a high-contrast medium, as it allows you to keep the main exposure on the short side, favoring shadows that would fall into a black abyss with a longer exposure, while still retaining some highlight tonality.
It's not all roses, however. Given the non-proportional density that flashing provides, the net result is that the highlights will end up rather flat (again assuming a positive working material). They will have tonality, but they will not necessarily differentiate very well. A compromise can be struck between differentiation and tone through careful 'dialing in' the flash exposure. Using this insight to answer your question about 'proper' flash exposure: there
is no straight answer to that question. It depends on what you're after. A desirable flash exposure can be the slightest nudge right up to the exposure threshold, or it can be a substantial bump right across this threshold to flatten out the bottom part of the HD curve. It depends on the effect you're after and the brightness range of the scene you're trying to capture.
Speaking in general, a suitable flash exposure is an exposure that gives no visible density if it's the only exposure the material receives. You can use a test strip approach to determine this. For something like using direct positive paper for capturing actual scenes, I would personally prefer to preflash the paper in the controlled conditions of the darkroom, before loading it into a film holder/camera. Trying to do this 'out in the field' complicates matters a bit especially if the material is not panchromatic, as the color distribution of the scene will impact the effectiveness of the preflash exposure. However, in principle, there's nothing wrong with using e.g. a diffusing filter in front of the lens or pinhole and preflashing that way (note that the diffusion is obviously a requirement; otherwise you're just stacking two regular exposures on top of each other). You'd still have to figure out the desirable preflash exposure, which will be dependent on the transmission of the diffusion filter, the recording material and the intended effect. So no straight answer.
In terms of approximating a ballpark figure, you could have a look at the curve of the material you use (which can be found in this case in
its datasheet) and then determine what degree of exposure would get you near the toe of the curve. Eyeballing the curve a bit, I'd say this would be some 2 stops below its normal exposure or so. Given the steepness of the curve, a crude approximation may not be the best approach and I'd still recommend doing some test strips to determine what works for you.
Hope this helps in any way. If you happen to do some comparative tests, it would be most useful if you could post them here for others to learn from. As you've noted, this paper isn't discussed a whole lot, so any real-world application information shared about it is welcome.
PS: apologies if you already knew all of this. It's not necessarily ground-breakingly new information; I realize this.
