[WIP] trigger github actions to figure out what our tests actually do.

[marsella]

This is a draft, please do not review yet.

Remember to remove pull request designation before merging.

Things to test before reviewing:

• [ ] what happens when the main testing fails?
• [ ] what happens when the customer watcher fails?
• [ ] what happens when the merchant watcher fails?

Other questions to explore (now or later):

• [ ] The test-all scenario runs a lot of tests, but doesn't designate expected outcomes. How can we add this designation?
• [ ] The test script does not look at intermediate state -- e.g. if the script does not fail, it assumes the test ran correctly. This means it misses classes of errors where, e.g., one party closes and the other doesn't notice and never updates their status to CLOSED.

[marsella]

Closing this - see #69 for the new testing plan and #361 for the first draft.