[BugFix] - PennyLane raises an informative error if running circuit(s) with dynamic tracers as wires

[PietropaoloFrisoni]

Context: Pennylane provides incorrect results or non-informative errors (depending on the circumstances) when working with @jax.jit and dynamic wires (that is, JAX tracers) at the same time. For example:

dev = qml.device("default.qubit")

@jax.jit
@qml.qnode(dev)
def circuit(input_wires):
    qml.Hadamard(input_wires[1])
    return qml.probs(wires=[0, 1])

circuit([0, 1])

Description of the Change: Pennylane raises an informative error in the validate_device_wires transform to inform the user that abstract wires are not currently supported. We cannot raise this error in the Wires class since this would be a problem for Catalyst. This solution should not (hopefully) cause issues in other repositories.

Benefits: This prevents non-informative errors and, most importantly, wrong results.

Possible Drawbacks: This change could potentially break existing code(s) that somehow used abstract wires and for some reason they worked fine.

Related GitHub Issues: #6380

Related ShortCut Stories: [sc-75756]

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 99.40%. Comparing base (e0f614c) to head (cd78015).
Report is 3 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #6405   +/-   ##
=======================================
  Coverage   99.40%   99.40%           
=======================================
  Files         447      447           
  Lines       42400    42418   +18     
=======================================
+ Hits        42146    42164   +18     
  Misses        254      254           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[JerryChen97]

lgtm!

[albi3ro]

Excellent choice of location for this fix 👍