Pyqrack problems with registers declared in for loop

[jon-wurtz]

When I try to interpret the code from our repeat until success example https://github.com/QuEraComputing/bloqade/blob/2e49ee75e79a9ed984a76aa14de4b609d69fd4e2/docs/digital/examples/repeat_until_success.py#L145

and replacing postselect_main with loop_main, running the example raises ValueError: qubit allocation exceeds memory, 1 qubits, 2 allocated

[weinbe58]

I don't think the current analysis will work unless you fold the entire program.

[weinbe58]

@Roger-luo this example is interesting because the dataflow analysis doesn't allow one to look inside the subroutine to see if the number of qubits is concrete or not. I think the analysis needs to keep track of the "largest" lattice element encountered.

[Roger-luo]

hmm, might related to QuEraComputing/kirin#217 or the ability of retriggering a sub-analysis within an analysis?