Rewrite for efficient execution / lpe transformation

[pjljvandelaar]

As a user, I would like to just write my models without considering performance impact.
The following two models are equivalent, yet their performance differs quite a lot

 ( p[A]() ||| q[B]() ||| r[C]() )
|[ A,B,C ]|
 ( k[A]() ||| l[B]() ||| m[C]() )

versus

 (p[A]() |[A]| k[A]())
|||
  (q[B]() |[B]| l[B]())
|||
  (r[C]() |[C]| m[C]())

Articles such as A graphical composition theorem for networks of LOTOS processes by T. Bolognesi and
Correctness Preserving Transformations for the Early Phases of Software Development by Tommaso Bolognesi, David De Frutos, Rom Langerak, and Diego Latella might provide more examples for rewrite rules.

Note I don't know for certain that fastest execution is also always fastest lpe transformation.
I would prefer that LPE can transform quickly, and the resulting lpe model is as fast as possible.