Package headers into pip wheel (#5734)

Summary:
Changes `CustomBuildPy` to take a list of source directories relative to ExecuTorch root and copies all headers recursively into pip wheel during packaging.

The destination is hardcoded to `executorch/include/executorch` in the pip wheel.

Pull Request resolved: #5734

Test Plan:
```bash
python setup.py bdist_wheel
```

This prints out the following lines:

```
creating pip-out/bdist.linux-x86_64/wheel/executorch/include/executorch/runtime
creating pip-out/bdist.linux-x86_64/wheel/executorch/include/executorch/runtime/core
copying pip-out/lib.linux-x86_64-cpython-311/executorch/include/executorch/runtime/core/array_ref.h -> pip-out/bdist.linux-x86_64/wheel/executorch/include/executorch/runtime/core
...
```
Then install the wheel:

```bash
pip install dist/executorch-0.5.0a0+52d5218-cp311-cp311-linux_x86_64.whl
```

And we can find the headers in site-packages

```
/data/users/larryliu/executorch/pip-out/lib.linux-x86_64-cpython-311/executorch/include/executorch/runtime/core/array_ref.h
```

Reviewed By: dbort

Differential Revision: D63561966

Pulled By: larryliu0820

fbshipit-source-id: 0c7d8983a89d11e62da006d361ca00b4a061d73c

Author: larryliu0820

setup.py

@@ -424,6 +424,25 @@ def run(self):
                 "devtools/bundled_program/serialize/scalar_type.fbs",
             ),
         ]
+        # Copy all the necessary headers into include/executorch/ so that they can
+        # be found in the pip package. This is the subset of headers that are
+        # essential for building custom ops extensions.
+        # TODO: Use cmake to gather the headers instead of hard-coding them here.
+        # For example: https://discourse.cmake.org/t/installing-headers-the-modern-
+        # way-regurgitated-and-revisited/3238/3
+        for include_dir in [
+            "runtime/core/",
+            "runtime/kernel/",
+            "runtime/platform/",
+            "extension/kernel_util/",
+            "extension/tensor/",
+            "extension/threadpool/",
+        ]:
+            src_list = Path(include_dir).rglob("*.h")
+            for src in src_list:
+                src_to_dst.append(
+                    (str(src), os.path.join("include/executorch", str(src)))
+                )
         for src, dst in src_to_dst:
             dst = os.path.join(dst_root, dst)