The path where headers are installed on Windows is not found (by default) by setuptools

- [X] I read [the conda-forge documentation](https://conda-forge.org/docs/user/introduction.html#how-can-i-install-packages-from-conda-forge) and could not find the solution for my problem there.

Issue:
On Linux, the conda xtensor-python package installs the headers at `$envprefix/include`, which is searched by default when building extensions with setuptools. This is good!
On Windows, however, the headers are installed at `$envprefix/Library/include`, which is not searched by setuptools by default (one needs to fiddle with `os.environ["CONDA_PREFIX"]` in `setup.py` to add the path). Instead, they could be installed at `$envprefix/include`, and things would work fine.

One can repro e.g. by running `python setup.py build` with the following setup.py:
```python
from setuptools import setup
from pybind11.setup_helpers import Pybind11Extension setup(
name="testpkg",
ext_modules=[Pybind11Extension("pkg", ["pkg.cpp"])],
)
```
and pkg.cpp, basically copied from xtensor-python's docs:
```c++
#include // Standard library import for std::accumulate
#include "pybind11/pybind11.h" // Pybind11 import to define Python bindings
#include "xtensor/xmath.hpp" // xtensor import for the C++ universal functions
#define FORCE_IMPORT_ARRAY // numpy C api loading
#include "xtensor-python/pyarray.hpp" // Numpy bindings

double sum_of_sines(xt::pyarray& m)
{
auto sines = xt::sin(m); // sines does not actually hold values.
return std::accumulate(sines.cbegin(), sines.cend(), 0.0);
}

PYBIND11_MODULE(pkg, m)
{
xt::import_numpy();
m.doc() = "Test module for xtensor python bindings";

m.def("sum_of_sines", sum_of_sines, "Sum the sines of the input values");
}
```

(Note that the same problem actually applies for xtensor itself, but I'm only opening it here.)

Environment (conda list):

Clean environment created with `conda create -n tmpenv -c conda-forge python=3.9 xtensor-python`.

```
$ conda list
# packages in environment at C:\Users\Antony\miniconda3\envs\tmpenv:
#
# Name Version Build Channel
ca-certificates 2021.5.30 h5b45459_0 conda-forge
certifi 2021.5.30 py39hcbf5309_0 conda-forge
intel-openmp 2021.2.0 h57928b3_616 conda-forge
libblas 3.9.0 9_mkl conda-forge
libcblas 3.9.0 9_mkl conda-forge
liblapack 3.9.0 9_mkl conda-forge
mkl 2021.2.0 hb70f87d_389 conda-forge
numpy 1.20.3 py39h6635163_1 conda-forge
openssl 1.1.1k h8ffe710_0 conda-forge
pip 21.1.2 pyhd8ed1ab_0 conda-forge
pybind11 2.6.2 py39h2e07f2f_0 conda-forge
pybind11-global 2.6.2 py39h2e07f2f_0 conda-forge
python 3.9.4 h7840368_0_cpython conda-forge
python_abi 3.9 1_cp39 conda-forge
setuptools 49.6.0 py39hcbf5309_3 conda-forge
sqlite 3.35.5 h8ffe710_0 conda-forge
tbb 2021.2.0 h2d74725_0 conda-forge
tzdata 2021a he74cb21_0 conda-forge
vc 14.2 hb210afc_4 conda-forge
vs2015_runtime 14.28.29325 h5e1d092_4 conda-forge
wheel 0.36.2 pyhd3deb0d_0 conda-forge
wincertstore 0.2 py39hcbf5309_1006 conda-forge
xtensor 0.23.10 h5362a0b_0 conda-forge
xtensor-python 0.25.2 py39h66837cb_0 conda-forge
xtl 0.7.2 h5362a0b_1 conda-forge
```

Details about conda and system ( conda info ):

```
$ conda info
active environment : tmpenv
active env location : C:\Users\Antony\miniconda3\envs\tmpenv
shell level : 3
user config file : C:\Users\Antony\.condarc
populated config files :
conda version : 4.9.2
conda-build version : 3.20.5
python version : 3.8.3.final.0
virtual packages : __win=0=0
__archspec=1=x86_64
base environment : C:\Users\Antony\miniconda3 (writable)
channel URLs : https://repo.anaconda.com/pkgs/main/win-64
https://repo.anaconda.com/pkgs/main/noarch
https://repo.anaconda.com/pkgs/r/win-64
https://repo.anaconda.com/pkgs/r/noarch
https://repo.anaconda.com/pkgs/msys2/win-64
https://repo.anaconda.com/pkgs/msys2/noarch
package cache : C:\Users\Antony\miniconda3\pkgs
C:\Users\Antony\.conda\pkgs
C:\Users\Antony\AppData\Local\conda\conda\pkgs
envs directories : C:\Users\Antony\miniconda3\envs
C:\Users\Antony\.conda\envs
C:\Users\Antony\AppData\Local\conda\conda\envs
platform : win-64
user-agent : conda/4.9.2 requests/2.23.0 CPython/3.8.3 Windows/10 Windows/10.0.19041
administrator : False
netrc file : None
offline mode : False
```

Commencez par la reproduction avec setup.py et l’exemple pkg.cpp, puis examinez comment le paquet conda pour Windows place les en-têtes de xtensor-python sous $envprefix/Library/include. Confirmez la modification en exécutant python setup.py build sous Windows et en vérifiant que setuptools trouve les en-têtes dans le chemin d’inclusion par défaut.

Rédigé par le modèle d'indexation à partir du texte de l'issue.