I’d be reluctant to remove the boss on the underside of the slide, for obvious functional reasons (let alone sacrilege!)
Adding tapped holes to the cross-slide saddle could work but would attract swarf, so would need plugging when not in use.
‘
The Beecroft method avoids those problems but the intermediate plate approach is structurally more sound. Its disadvantage (reduced centre-height) can be alleviated to some extent by using a thinner plate with a stiffening rib along each side that drapes down the side of the cross-slide: e.g. from 3/8 down to 1/4 or even 3/16 inch. Bore the centre to take the top-slide boss.
In that regard, Mark’s modification to the Beecroft design is the best of these three options.
.
The problem of the handles fouling at extreme angles might be solved (I have not tried this) by locking the cross-slide at a suitable position then removing its handle and screw. You might need use some deft calculations if working to a step rather than full-diameter facing.
.
I like Michael’s copy-turning suggestion for the reason Mike Poole advances. To give the positive tracking Mr. Poole queries, a straight conical-facing template could be a round bar taking a simple linear-bearing follower (a plain brass bush should be ample). For curves and compound shapes, use a pin or roller follower running in a channel in a plate or flat bar.