Use --no-build-isolation in tests to avoid using Internet

Add `--no-build-isolation` to the pip invocation in `pip_editable_parts`
fixture and a few tests, to avoid downloading `setuptools` wheel from
PyPI.  This change makes it possible again to run the vast majority of
pip tests (with the exception of 3 tests) offline.

Fixes #12786.

Author: mgorny

tests/conftest.py

@@ -395,6 +395,7 @@ def pip_editable_parts(
             "-m",
             "pip",
             "install",
+            "--no-build-isolation",
             "--target",
             pip_self_install_path,
             "-e",

tests/functional/test_config_settings.py

@@ -118,6 +118,7 @@ def test_config_settings_implies_pep517(
     )
     result = script.pip(
         "wheel",
+        "--no-build-isolation",
         "--config-settings",
         "FOO=Hello",
         pkg_path,

tests/functional/test_install.py

@@ -685,7 +685,9 @@ def test_link_hash_in_dep_fails_require_hashes(
     # Build a wheel for pkga and compute its hash.
     wheelhouse = tmp_path / "wheehouse"
     wheelhouse.mkdir()
-    script.pip("wheel", "--no-deps", "-w", wheelhouse, project_path)
+    script.pip(
+        "wheel", "--no-build-isolation", "--no-deps", "-w", wheelhouse, project_path
+    )
     digest = hashlib.sha256(
         wheelhouse.joinpath("pkga-1.0-py3-none-any.whl").read_bytes()
     ).hexdigest()
@@ -903,7 +905,14 @@ def test_editable_install__local_dir_setup_requires_with_pyproject(
         "setup(name='dummy', setup_requires=['simplewheel'])\n"
     )
 
-    script.pip("install", "--find-links", shared_data.find_links, "-e", local_dir)
+    script.pip(
+        "install",
+        "--no-build-isolation",
+        "--find-links",
+        shared_data.find_links,
+        "-e",
+        local_dir,
+    )
 
 
 def test_install_pre__setup_requires_with_pyproject(

tests/functional/test_self_update.py

@@ -11,12 +11,12 @@ def test_self_update_editable(script: Any, pip_src: Any) -> None:
     # Step 1. Install pip as non-editable. This is expected to succeed as
     # the existing pip in the environment is installed in editable mode, so
     # it only places a .pth file in the environment.
-    proc = script.pip("install", pip_src)
+    proc = script.pip("install", "--no-build-isolation", pip_src)
     assert proc.returncode == 0
     # Step 2. Using the pip we just installed, install pip *again*, but
     # in editable mode. This could fail, as we'll need to uninstall the running
     # pip in order to install the new copy, and uninstalling pip while it's
     # running could fail. This test is specifically to ensure that doesn't
     # happen...
-    proc = script.pip("install", "-e", pip_src)
+    proc = script.pip("install", "--no-build-isolation", "-e", pip_src)
     assert proc.returncode == 0