Numpy types are allowed in kernel signatures but not created as such during return

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.
• If possible, make a PR with a failing test to give us a starting point to work on!

Describe the bug

Even though we allow to use numpy integers and floating point types in kernel signatures, the returned value is always a standard type. Kernels respect the requested bitwidths passed for arguments, but return values are not handled properly to respect it.

Steps to reproduce the bug

There are some test cases in the code with a link to this issue posted.

Expected behavior

We should create the appropriate numpy values.

Is this a regression? If it is, put the last known working version (or commit) here.

Not a regression

Environment

• CUDA-Q version: latest main (as of https://github.com/NVIDIA/cuda-quantum/pull/3489)
• Python version: N/A
• C++ compiler: N/A
• Operating system: N/A

Suggestions

No response

Locate the test cases in the code that reference issue 3524, then trace the kernel return-value path for NumPy integer and floating-point signatures. Compare it with argument handling and verify that returned values preserve the requested NumPy type and bitwidth; done means the relevant tests pass for the supported types.