". . . the C (C for collar) notation . . . cannot cover elongated, more complex knots, where we have the working end coming into play with the tail . . ."
Well, for those knots, all SPATIAL regions that come into existence as the knot is made have no associating syntactic representation; but, using the T specification, all crossings that occur BELOW the S region are officially in the T region. See? Most of the knot (except possibly for the crossing that touches the standing region) can be considered to be in the T region, since it was formed strictly by the tail's moving over and under itself. With the T specification, your notation will be able to accommodate the overs and unders of a tail's crossing over and under itself, for as long a sequence as is necessary to finish a knot. Since a practical knot can be only so big, the "sequence" will automatically be limited to a practical length.
Considering the possibilities of sequence, selection, iteration, and recursion, I'd say that sequence is the best choice!
". . . is it more natural for most people to tie the overhand knot in such a way that the tail emerges above the L bight (and the standing end below the H bight)?"
Yes, the notation needs to be able to handle all of the possibilities equally well.
When making a Bowstring Loop, I go under and over and under; when making a Butterfly End Loop, I go over and under and over. As if that weren't complicated enough, my father was left-handed, I am right-handed, and I use both hands when playing piano!
JCS