Grabbing the body of the shunt (rather than the carabiner) will cause the shunt to slide, and in this way the Shunt is subject to the same panic problems as a single-stop descender.
You may find, as NZcaver suggests, that it is not so easy to unweight the shunt. Without a separate ascender, you'll probably be unable to do it. Even with an ascender, you'll need to lock off your descender. This difficulty in transferring weight off it is something you may encounter with it below the descender, but it's something you'll
definitely encounter with it above the descender.
I believe Hank Moon is suggesting
this technique for placing the shunt or friction hitch below the descender. This technique has the advantage that it interferes with your ability to rappel normally the least. It is also at least as safe as rappelling normally. (Though I would take issue with the climber in the picture clipping into the carabiner that it is on just one of the pitons. Not only does he take a shockload onto
slingsif that piton blows out, but he probably shockloads the carabiner along the short axis. He should instead clip into the power point of the anchor system, and the only disadvantage is that he will then start his rappel slightly lower.
The disadvantage of doing it this way is that normal rappelling consists of a firm grip with the brake hand, and above that, a light grip with the prusik-minding hand. Slowing down consists of a harder grip with the brake hand and, above that, the same light grip with the prusik-minding hand. If you were to lose your brake hand, your instinctive response might not be to either let go or grab hard on the prusik above, but rather to maintain your light prusik-mind behavior. Certainly, once you realized you were falling, you would either wisely let go of it, or foolishly grab it, and
either action would increase tension on the brake end of the rope and bring the rappel under control. But by that time, the friction hitch could be sliding fast enough to melt when it catches. (I don't know what a shunt does when it is released into a fast-moving rope. However, if you instinctively grab the body of the shunt in this situation, you keep falling.)
I would consider the method which (I am saying) Hank is advocating outright dangerous and extremely unlikely to succeed at saving your life
when done with a shunt. With a friction hitch, I think it would be more likely to succeed...but it would be interested to know if there is real-world evidence to support that.
Anchoring a friction hitch to the leg loop has the disadvantages Hank mentions, that (1) it will be minded or munched, and consequently most likely fail to catch, if it is ever able to reach the descender, and (2) leg loops are not generally designed to be loaded in this way. But if you do succeed at getting the geometry right and you rappel by holding your brake hand on the hitch, then if you grab the hitch harder in a fall, you're doing exactly what you're supposed to do--increasing your hold on the brake end of the rope. (Which is the same effect with the Petzl method.) On the other hand, if you were to let go, it would grab by itself (also the same effect with the Petzl method). Unlike with the Petzl method though, there is no normal behavior that could extend in a panic situation to minding the friction hitch without applying a firm hold on the brake end of the rope. Nobody applies a moderately light hold with the brake hand while rappelling, as a result of panic.
On the issue of how to extend a point of attachment to a cave rope, I recommend a cowstail made from dynamic single rope (the kind climbers would use by itself without another rope, as the primary shock-absorbing component of their fall arrest system). For the most shock absorption, you should probably use a figure-eight on a bight on an end where you want a loop of fixed size, and a barrel knot on an end where you want the loop to contract around a connector and hold it in place. There's a discussion of these matters in
this long thread.