`np.recarray` is not properly converted

**Affects:** PythonCall

**Describe the bug**
It seems that `pyconvert(Any, ::)` incorrectly assumes that a `recarray` can be wrapped as a `PyArray`.

```
julia> np = pyimport("numpy");

julia> arr = np.recarray((2,2), dtype = @py([("A", "O"), ("B", "O")]))
Python:
rec.array([[(None, None), (None, None)],
[(None, None), (None, None)]],
dtype=[('A', 'O'), ('B', 'O')])

julia> pyconvert(Any, arr)
2×2 PyArray{NamedTuple{(:A, :B), Tuple{PythonCall.Wrap.UnsafePyObject, PythonCall.Wrap.UnsafePyObject}}, 2}:
(A = UnsafePyObject(Ptr{PyObject} @0x00007ffc75ba4830), B = UnsafePyObject(Ptr{PyObject} @0x00007ffc75ba4830)) (A = UnsafePyObject(Ptr{PyObject} @0x00007ffc75ba4830), B = UnsafePyObject(Ptr{PyObject} @0x00007ffc75ba4830))
(A = UnsafePyObject(Ptr{PyObject} @0x00007ffc75ba4830), B = UnsafePyObject(Ptr{PyObject} @0x00007ffc75ba4830)) (A = UnsafePyObject(Ptr{PyObject} @0x00007ffc75ba4830), B = UnsafePyObject(Ptr{PyObject} @0x00007ffc75ba4830))
```

A field name can not be accessed after conversion.
```
julia> arr.A
Python:
array([[None, None],
[None, None]], dtype=object)

julia> pyconvert(Any, arr).A
ERROR: type PyArray has no field A
Stacktrace:
[1] getproperty(x::PyArray{NamedTuple{(:A, :B), Tuple{PythonCall.Wrap.UnsafePyObject, PythonCall.Wrap.UnsafePyObject}}, 2, true, false, NamedTuple{(:A, :B), Tuple{PythonCall.Wrap.UnsafePyObject, PythonCall.Wrap.UnsafePyObject}}}, f::Symbol)
@ Base .\Base.jl:37
[2] top-level scope
@ REPL[11]:1

```

If indexing before field access, it works, but it does not return a usable wrapper.
```
julia> pyconvert(Any, arr)[1]
(A = PythonCall.Wrap.UnsafePyObject(Ptr{PythonCall.C.PyObject} @0x00007ffc75ba4830), B = PythonCall.Wrap.UnsafePyObject(Ptr{PythonCall.C.PyObject} @0x00007ffc75ba4830))

julia> pyconvert(Any, arr)[1].A
PythonCall.Wrap.UnsafePyObject(Ptr{PythonCall.C.PyObject} @0x00007ffc75ba4830)
```

I think the expected behavior should be `pyconvert(Any, ::)` returning something equivalent to a [`StructArray`](https://github.com/JuliaArrays/StructArrays.jl), i.e. `foo[1].A` and `foo.A[1]` are equivalent.

**Environment:**
Julia v1.9.3
PythonCall v0.9.23

Start at the pyconvert handling for numpy recarray values and compare it with the reported conversion output. Reproduce the example with np.recarray, then make the converted result support both foo[1].A and foo.A[1] as described, using StructArray as the stated behavioral reference.

索引モデルが issue の本文から書いたものです。