Add more data-driven tests. #3601

Dont even ask who failed to pass these elsewhere...

Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/c32a486f4b2253900749acd0d42efae3319b98ed for binaries and test log. This test script verifies pulls every time they are updated. It, however, dies sometimes and fails to test properly. If you are waiting on a test, please check timestamps to verify that the test.log is moving at http://jenkins.bluematt.me/pull-tester/current/ Contact BlueMatt on freenode if something looks broken.

Was this discussed in IRC and ACK'd there? No objections to the PR content, but "add tests" is something that needs review and multiple ACKs like anything else... Otherwise we risk setting in stone unreviewed behavior. These tests are copied into other, non-bitcoind projects as canonical reference tests, so it is critical to make sure these remain correct.

if you can represent the behavior using valid scripts, then, whether it's expected behavior or not, it defines the protocol. (The obvious exception being some script behavior which represents a potential security risk, which, I would, hope would not be submitted in a pull request)

Ideally, this sort of test should be verified to succeed in multiple (including old) releases. If they do (and the test checks something that is network-visible behaviour), it is clearly a network rule (even if the behaviour is unintentional) and it should be something that other implementations verify. If they don't, we likely have a big problem already.

Ok, I thought adding more tests cannot hurt.

Revert this for now?

@sipa ideally we'd have lots more testing in lots more ways :p @laanjw No, I think the discussion here is more general and not specific to these test cases.

Posthumous ACK. No need to revert this.