Unzip CI results, fix update script

1. Unzip expected CI results, update scripts accordingly.
2. Fix results for 'heap' and 'lists' test-sets.
3. Fix updating results: allow script to change expected outcomes:

When there are multiple outcomes, don't fail. Use the last (new) outcome, but warn the user just in case.

Do fail if there are more than two outcomes (presumably there should be only one old and only one new outcome).

If there were multiple new outcomes across the new results, it is undetermined which one would be written as the expected one.

Ticks some checks in #28.