Now the rw guys may correct me there but I think that Mike Ray describes a sim only procedure there, using the wrong self-build LNAV waypoints for arcs.
The use of raw data is true, but the real world fellows in a 757 or 767 would either have a FMC database to describe that arc in full detail or they are unable to use FMC guidance there, at that procedure, using a DME arc.
Of course, this statement may well depend on the actual tolerance allowed for that special procedure, in that country, at that airport. If you're allowed to deviate more, the edged "arc description" may well be ok for this.

Needs some Lourification I think.  :D

Mike Ray may describe the stuff for the same reason why e. g. Michael does. It's the only way to "cheat" your sim FMC to fly close to an arc.
A real FMC would have this for example.

Quote:

Radius to fix legs — Implemented the ARINC 424 leg type that provided a fixed radius ground path (similar to a DME arc but without the required navaid).

Since no current sim FMC can understand such descriptions, the 'describe an arc with lots of waypoints' method is indeed common.
You can spot the advantage of the FMC based approach too, it doesn't actually need an installed navaid on the ground while still being able to precisely track the way as given in the approach description.

Regarding the shortening of routes, I've seen other dev's FMCs doing this too, more or less severe, so even if they are using an "overfly flag" (which is a good point by the way), they seem to struggle from some logical problems there.
Now I can't measure the actual "rounding capability" at my addons, but, so far, they all need some precaution from the user's side to avoid funny looking approaches when not using easy "straight in" all the time.