tracking issue: Control flow rewrites / flattening

[aborgna-q]

Having control flow primitives so at-hand brings us problems when running optimisation designed for simple flat quantum circuits.
Since now it's easier for a user to write a guppy loop to apply gates to all qubits than to add them one-by-one, we have to be able to deal with such higher level constructions if we are to match the optimisation performance of the legacy pytket compiler.

Although this is an open goal, we should start the process by implementing some straightforward optimisations and rewrites.