Fix bug in exp_pauli qubit concatenation

[huaweil-nv]

Description

Fix a Python operator precedence bug in PyKernel.exp_pauli() where individual qubits (RefType) are silently dropped when a quantum register (VeqType) is also provided.
The + operator has higher precedence than the ternary if/else expression in Python, so:

[quantumVal] if quantumVal is not None else [] + qubitsList

is parsed as:

[quantumVal] if quantumVal is not None else ([] + qubitsList)

This means when quantumVal is not None, the result is [quantumVal] alone — qubitsList is discarded entirely. The fix adds parentheses to ensure qubitsList is always concatenated:

([quantumVal] if quantumVal is not None else []) + qubitsList

How to reproduce:

import cudaq

kernel = cudaq.make_kernel()
qreg = kernel.qalloc(2)
q_extra = kernel.qalloc()
kernel.exp_pauli(1.0, qreg, q_extra, 'XXX')

print(kernel)

Before fix: The generated IR shows q_extra allocated but never used. exp_pauli operates on veq<2> (2 qubits) despite the 3-qubit Pauli word "XXX":

%0 = quake.alloca !quake.veq<2>
%1 = quake.alloca !quake.ref
quake.exp_pauli (%cst) %0 to "XXX" : (f64, !quake.veq<2>) -> ()

After fix: quake.concat correctly merges the register and the individual qubit into veq<3>:

%0 = quake.alloca !quake.veq<2>
%1 = quake.alloca !quake.ref
%2 = quake.concat %0, %1 : (!quake.veq<2>, !quake.ref) -> !quake.veq<3>
quake.exp_pauli (%cst) %2 to "XXX" : (f64, !quake.veq<3>) -> ()

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[sacpis]

LGTM. Thansk @huaweil-nv.

[sacpis]

/ok to test c06e3be

Command Bot: Processing...

[github-actions]

CUDA Quantum Docs Bot: A preview of the documentation can be found here.