[python] exp_pauli lacks the individual-qubit operand form available in C++

[wsttiger]

Required prerequisites

• Consult the security policy. If reporting a security vulnerability, do not report the bug using this form. Use the process described in the policy to report the issue.
• Make sure you've read the documentation. Your issue may be addressed there.
• Search the issue tracker to verify that this hasn't already been reported. +1 or comment there if it has.

Describe the feature

C++ kernels can apply a Pauli-word exponential to explicitly listed
qubits: exp_pauli(theta, "YX", qubits[a], qubits[b]). The Python kernel
builder only accepts exp_pauli(angle, register, word) where the word
spans the whole register; the individual-qubit form is rejected with "too
many values".

import cudaq
cudaq.set_target("qpp-cpu")

@cudaq.kernel
def kernel(theta: float):
    q = cudaq.qvector(3)
    exp_pauli(theta, "YX", q[0], q[1])   # error: too many values

# Works today: contiguous runtime slice
@cudaq.kernel
def kernel2(theta: float, a: int):
    q = cudaq.qvector(4)
    exp_pauli(theta, q[a:a + 2], "YX")   # OK, runtime offset supported

For contiguous pairs the runtime-slice form works (verified, including
runtime offsets), but non-contiguous operand sets require padding the word
with identities to full register width on the host. Parity with the C++
operand form would make C++-to-Python kernel ports mechanical.

Environment

• CUDA-Q built from source at commit 0be565550f4c23affdcbed9e4eaec38d2d0915e6
• Python 3.11.x, target qpp-cpu (library mode)
• AlmaLinux 8.10, x86_64

All reproducers below were executed and verified against this build.