DAST expectations should use a single report file
Following my discussion with @dappelt, I find the current expectations files suited for tests, but not for humans. They're actually chunks of the final report and used as a comparison with extracted portions (using jq).
We used this technique because the tests are running inside a docker:stable image, which has only sh (and no bash). Hence, it's harder to diff two jq extracts, like we could do with bash: diff <(jq -S . A.json) <(jq -S . B.json) (this won't work on sh).
This issue is to discuss the balance between having a complete report file vs having temporary files (extracted with jq) in tests.