While writing WT-12506 I realised the output produced by s_clang_format is hard to understand. This ticket is to clean up the output and make it easier to understand

The way the script works is:

• Run clang_scan
• Capture the output from clang_scan, filter it to only report files and warnings (removing context such as line numbers)
• Diff the filtered output with s_clang_scan.diff and fail if there is any delta
• When there's a failure also dump the unfiltered clang_scan output to help with debugging

This produces two different fail modes

1. If there are suppressed errors in s_clang_scan.diff which aren't present in the clang_scan output then we just need to remove those lines from clang_scan.diff. This is not a real error and s_clang_scan.diff needs to be updated
2. If there are errors in the clang_scan output which aren't suppressed by s_clang_scan.diff then we have an issue that needs to be addressed.

Currently in both cases we'll report the diff and dump the unfiltered clang_scan output. This causes confusion as developers see the unfiltered output and assume they need to address errors which are already suppressed by s_clan_scan.diff.

We should update the script to do the following:

• In case 1) only report that lines need to be removed from s_clang_scan.diff
• In case 2) Find a way to still report the unfiltered output, but make it clear that the only warnings that need to be addressed are the ones from the diff output reported under the

  =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
  scan-build output differs from expected:
  <<<<<<<<<< Expected >>>>>>>>>> Current
  =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= 

  heading