Did we understand what caused #2116? This feels like a (legit) workaround but I feel I didn't understand the root cause of it.

I infer that the reason is that we don't have a way to tell which .md files are more relevant for the rendered link, correct?

The root cause behind #2116 is that triagebot currently looks at the list of modified files from the PR and shows the first file to match one of the rules under trigger_files.

This is quite naïve from triagebot as it implicitly relies on the assumption that the most relevant modified file is always the first one in the PR, which it obviously isn't.

As a better metric this PR proposes that we use the total number of lines changed which should eliminate the index files (like SUMMARY.md or index.md) where only small changes are made compared to the rest from being linked in priority.