[llvm] [workflows] Split pr-code-format into two parts to make it more secure (PR #78216)

[Aiden Grossman via llvm-commits]

================
@@ -44,6 +44,7 @@ class FormatArgs:
     token: str = None
     verbose: bool = True
     issue_number: int = 0
+    write_comment_to_file: bool = False
----------------
boomanaiden154 wrote:

Given that the CI exists, I think that functionality would be pretty redundant. The only reason I've run it locally in the past and had it comment on a PR was for testing, which should be quite a bit easier now that it spits everything out into a file rather than needing to comment directly.

https://github.com/llvm/llvm-project/pull/78216