Fix shebang test to skip __init__/test files, fix forumdl cchardet build failure

1. Shebang test: skip executable bit + shebang checks for __init__.py,
   test files, and utils — these aren't meant to be run as scripts.

2. Forumdl test: add pip overrides matching the crawl hook — install
   with --no-deps and chardet==5.2.0 instead of cchardet, which can't
   compile on Python 3.12+ (longintrepr.h removed).

https://claude.ai/code/session_01XyzPfEB3qopffFuRyouaTa

Author: claude

abx_plugins/plugins/forumdl/tests/test_forumdl.py

@@ -58,6 +58,26 @@ def get_forumdl_binary_path() -> str | None:
     binary = Binary(
         name="forum-dl",
         binproviders=[PipProvider(), EnvProvider()],
+        overrides={
+            "pip": {
+                "install_args": [
+                    "--no-deps",
+                    "--prefer-binary",
+                    "forum-dl",
+                    "chardet==5.2.0",
+                    "beautifulsoup4",
+                    "soupsieve",
+                    "lxml",
+                    "requests",
+                    "urllib3",
+                    "tenacity",
+                    "python-dateutil",
+                    "six",
+                    "html2text",
+                    "warcio",
+                ]
+            }
+        },
     ).load_or_install()
     if binary and binary.abspath:
         _forumdl_binary_path = str(binary.abspath)

tests/test_plugin_executables_have_shebang.py

@@ -32,11 +32,15 @@ def test_all_plugin_scripts_are_executable_and_have_shebang() -> None:
 
     for script_path in _iter_plugin_scripts():
         rel_path = script_path.relative_to(REPO_ROOT)
+
+        if not _requires_shebang(script_path):
+            continue
+
         mode = script_path.stat().st_mode
         if not mode & (stat.S_IXUSR | stat.S_IXGRP | stat.S_IXOTH):
             failures.append(f"{rel_path}: missing executable bit")
 
-        if _requires_shebang(script_path):
+        if True:
             first_line = script_path.read_text(
                 encoding="utf-8", errors="ignore"
             ).splitlines()