github: scripts: fix commit prefix extraction capture in linter

  Update commit prefix regex handling in `.github/scripts/commit_prefix_check.py`
  so subject prefix extraction returns only the prefix token (e.g. `build:`)
  instead of including trailing content.

  Changes:
  - make `PREFIX_RE` capture just the prefix in group 1:
    - from: `^[a-z0-9_]+:\s+\S`
    - to:   `^([a-z0-9_]+:)\s+\S`
  - update subject prefix extraction to use `match.group(1)`

  Why:
  - previous extraction used `.group()` and could include extra matched text,
    causing incorrect prefix comparisons during validation.

  Notes:
  - `detect_bad_squash()` uses `PREFIX_RE.match(...)` only as a boolean;
    behavior remains unchanged there.

Signed-off-by: Eduardo Silva <[email protected]>

Author: edsiper

.github/scripts/commit_prefix_check.py

@@ -20,7 +20,7 @@
 repo = Repo(".")
 
 # Regex patterns
-PREFIX_RE = re.compile(r"^[a-z0-9_]+:\s+\S", re.IGNORECASE)
+PREFIX_RE = re.compile(r"^([a-z0-9_]+:)\s+\S", re.IGNORECASE)
 SIGNED_OFF_RE = re.compile(r"Signed-off-by:", re.IGNORECASE)
 FENCED_BLOCK_RE = re.compile(
     r"""
@@ -176,7 +176,7 @@ def validate_commit(commit):
     if not subject_prefix_match:
         return False, f"Missing prefix in commit subject: '{first_line}'"
 
-    subject_prefix = subject_prefix_match.group()
+    subject_prefix = subject_prefix_match.group(1)
 
     # Run squash detection (but ignore multi-signoff errors)
     bad_squash, reason = detect_bad_squash(body)