Our API can't represent non-UTF-8 content

For example: https://gitlab.com/api/v4/projects/gitlab-org%2fgitlab-test/repository/commits/79faa7bfb522480533f4d60368d8f6ecc1e03d9b/diff

You can see the patch at https://gitlab.com/gitlab-org/gitlab-test/commit/79faa7bfb522480533f4d60368d8f6ecc1e03d9b.patch

The problem here is that the API is returning data in JSON, and this diff contains bytes that aren't valid UTF-8. We should probably replace these with a placeholder, like @godfat suggested in the related issue https://gitlab.com/gitlab-org/gitlab-ce/issues/35098#note_35029239.