Nested C++ QPU kernel does not accept std::vector parameters anymore

[bebora]

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

nvq++ correctly compiles a quantum kernel with a std::vector parameter. However, if this std::vector is then passed to another kernel inside the original one, compilation fails.

Steps to reproduce the bug

Create vect.cpp:

#include <cudaq.h>
#include <vector>

__qpu__ void ApplyMaskedX(cudaq::qview<> qc, std::vector<int> mask)
{
    int s = qc.size();
    for (int i = 0; i < s; i++)
    {
        if (mask[i] == 1)
        {
            x(qc[i]);
        }
    }
}

struct RunKernel
{
    __qpu__ auto operator()(std::vector<int> mask)
    {
        cudaq::qvector qc(mask.size());
        ApplyMaskedX(qc, mask);
        mz(qc);
    }
};

int main()
{
    std::vector<int> mask = {1, 0, 1, 0, 0};
    auto res = cudaq::sample(1000, RunKernel{}, mask);
    res.dump();
}

Compile it:

$ nvq++ vect.cpp 
loc("vect.cpp":21:26):error: /cuda-quantum/lib/Frontend/nvqpp/ConvertExpr.cpp:2950: not yet implemented: C++ constructor (non-default)
vect.cpp:21:26: error: C++ constructor (non-default) is not yet supported
        ApplyMaskedX(qc, mask);
                         ^

The code compiles and runs if the innermost kernel is inlined inside the outer one as follows:

// includes...
struct RunKernel
{
    __qpu__ auto operator()(std::vector<int> mask)
    {
        cudaq::qvector qc(mask.size());
        int s = qc.size();
        for (int i = 0; i < s; i++)
        {
            if (mask[i] == 1)
            {
                x(qc[i]);
            }
        }
        mz(qc);
    }
};
// main...

Expected behavior

I expect to be able to pass outer std::vector parameters to sub-kernels to write composable functions.

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

0.9.0

Environment

• CUDA-Q version: 0.9.1 (pre-built binaries and cu12-0.9.1 Docker image)
• Operating system: Ubuntu 22.04.5 LTS

Suggestions

No response