QA Tests to protect Blockbook interoperability

[placer14]

In order to protect expected behavior between ob-go and the multiwallet API, we want to create a set of tests which will exercise the actual scenarios through which the API is expected to normally perform on a coin-by-coin basis. The tests should be implemented as an independent suite of tests which are exercised against a live blockbook API.

The test creators should take care that ob-go assurances aren't leaking into the tests and should only be assured when required to test a specific scenario as it relates to the blockbook API. (For example, we assert a node is offline to ensure the production of a 1-of-2 transaction and don't care as much that the node was offline so much as we care that it allowed the 1-of-2 transaction to take place. This is important from the perspective of naming functions and variables as part of the description of the test. In this case, I would prefer a function called AssertNodePreparedForOneOfTwoMultisig() instead of AssertNodeOffline().

Requirements:

• Tests should separate tests by coin and then by scenario
• Tests should allow individual coin support to be exercised independently against a specific API endpoint
• Each test should isolate the scenario to a specific behavior and not overload a test with asserting multiple behaviors at once (each asserted behavior should independently fail)

Scenarios:

• Fund and Get Balance
• Get block height info
• Get transaction by hash
• Send direct payment
• Send 1-of-2 payment
• Send 2-of-3 payment
• Test sweep wallet
• TBD...

This will require the builder MVP: OpenBazaar/mason#1

[placer14]

I want to investigate whether we can use regtest BB servers in these tests to better manage making and verifying payments. (See OpenBazaar/multiwallet#76)

On further investigation, it appears the build scripts for blockbook involve complex build, deb installation, and docker orchestration. This might be a tall order for the short term, so we will not try to create a deterministic blockbook instance to test again. Instead, the test script will check that the wallet has sufficient funds before executing tests dependent on coin use.