Change 2-of-2 integration test to 2-of-3

[rubensayshi]

add failing unittest to prove the error I'm trying to fix in #369

[dcousens]

Disregard previous comments, accidentally read this as 3 of 3, not 2 of 3.

[dcousens]

@rubensayshi I have added c79fecf to show case this test without the need for it to be an integration test.

What is shown there is the expected txHex, it verifies that no OP_0 should be present in the scriptSig (for a buildComplete); and as such, before bcf8d01 was a failing test.

Please ACK this and I'll close this PR. :)

edit: moved changes into a PR, see #372

[rubensayshi]

hmm, I changed the 2of2 integration test to 2of3, I think that's a better test case since it has a greater chance of failing than a 2of2 (for various reasons)

[coveralls]

Coverage remained the same at 99.24% when pulling c52420a on blocktrail:2of3-failing-test into b132714 on bitcoinjs:master.

[dcousens]

@rubensayshi I'm OK with this change, but I don't think it covers any less or more than 2-of-2, it simply covers the other case (fully signed vs partially signed).

However this does show case a better example of multisig to the typical user, so I'll update the README example after to merge.

[dcousens]

@rubensayshi to follow that line of thought, I changed it to a less traditional 2-of-4 which would better illustrate some parts of the API and provide a larger contrast to the example above it, the 2-of-2.

Naturally, it also tests something less common which is a bonus I guess.

[rubensayshi]

awesome!