scripts/checkpatch.pl: remove bogus patch prefix warning

Remove the 'patch prefix exists, appears to be a -p0 patch' warning
entirely as it is fundamentally flawed and can only produce false
positives.

Sometimes I create test files with names 'a' and 'b', and then get
surprised seeing this warning. It was not easy to understand where it
comes from.

How it works:
1. It extracts prefixes (a/, b/) from standard diff output
2. Checks if files/directories with these names exist in the project
   root
3. Warns if they exist, claiming it's a '-p0 patch' issue

This logic is wrong because:
- Standard diff/patch tools always use a/ and b/ prefixes by default
- The existence of files named 'a' or 'b' in the working directory is
  completely unrelated to patch format
- The working directory state may not correspond to the patch content
  (different commits, branches, etc.)
- In QEMU project, there are no single-letter files/directories in root,
  so this check can only generate false positives

The correct way to detect -p0 patches would be to analyze the path
format within the patch itself (e.g., absolute paths or paths without
prefixes), not check filesystem state.

So, let's finally drop it.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Link: https://lore.kernel.org/r/20251030201319.858480-1-vsementsov@yandex-team.ru
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

scripts/checkpatch.pl

@@ -1741,13 +1741,7 @@ sub process {
 			}
 		} elsif ($line =~ /^\+\+\+\s+(\S+)/) {
 			$realfile = $1;
-			$realfile =~ s@^([^/]*)/@@ if (!$file);
-
-			$p1_prefix = $1;
-			if (!$file && $tree && $p1_prefix ne '' &&
-			    -e "$root/$p1_prefix") {
-				WARN("patch prefix '$p1_prefix' exists, appears to be a -p0 patch\n");
-			}
+			$realfile =~ s@^[^/]*/@@  if (!$file);
 
 			if (defined $fileinfo && !$fileinfo->{isgit}) {
 				$fileinfo->{lineend} = $oldhere;