Run f2py's meson backend with the same python that ran f2py

On Debian, we sometimes have multiple Python 3 versions in the archive.
When using f2py3.X, I'd expect meson to run with the same python, not
the default python.

Not sure how far up the stack this should come from, I kept it simple
for this PR, if someone has better ideas, they can direct me.

numpy/f2py/_backends/_meson.py

@@ -4,6 +4,7 @@
 import errno
 import shutil
 import subprocess
+import sys
 from pathlib import Path
 
 from ._backend import Backend
@@ -27,6 +28,7 @@ def __init__(
         linker_args: list[str],
         c_args: list[str],
         build_type: str,
+        python_exe: str,
     ):
         self.modulename = modulename
         self.build_template_path = (
@@ -59,6 +61,7 @@ def meson_build_template(self) -> str:
     def initialize_template(self) -> None:
         self.substitutions["modulename"] = self.modulename
         self.substitutions["buildtype"] = self.build_type
+        self.substitutions["python"] = self.python_exe
 
     def sources_substitution(self) -> None:
         indent = " " * 21
@@ -138,6 +141,7 @@ def write_meson_build(self, build_dir: Path) -> None:
             self.flib_flags,
             self.fc_flags,
             self.build_type,
+            sys.executable,
         )
         src = meson_template.generate_meson_build()
         Path(build_dir).mkdir(parents=True, exist_ok=True)

numpy/f2py/_backends/meson.build.template

@@ -8,7 +8,7 @@ project('${modulename}',
                           ])
 fc = meson.get_compiler('fortran')
 
-py = import('python').find_installation(pure: false)
+py = import('python').find_installation('${python}', pure: false)
 py_dep = py.dependency()
 
 incdir_numpy = run_command(py,