TTStacks usefull at all?
One the one hand the advantage of the lazy evaluation of TTStacks only shows up for more than one Operator and increases with the number of the operators.
On the other hand the (current) lazy evaluation forces the contraction of all nodes in a row, leading to an escalation of the ranks. In particular the ranks near the ends of the TTTensor usually largely exceed the maximal possible rank.
Directly applying the contraction for each operator would allow recanonicalization and thereby a significant rank reduction at the ends and possibly also at the other positions (at least some minimal tests suggest so).
There it is doubtful whether the use of TTStacks is useful.