Peer DID Interop tests - Peer DID Transition

[nodlesh]

This PR aims to partially complete #752. This is a draft PR until all ACA-Py tasks are completed below. AFJ tasks and other backchannel updates will be broken out into separate tickets if a ticket doesn't exist for it yet.

• Create Feature File and test definitions for Peer DIDs.
• In this PR, there is also some cleanup of the format of the json output in the ACA-Py Backchannel for readability.
• Add handling to restart an agent with different arguments if the Given "n" agents step contains startup args.
• Get start API call to accept any startup options, including --emit-peer-did-#
• In the ACA-Py Backchannel, make sure args passed in through the start API take precedence over the default args in the backchannel.
• After the feature is finished, change the reset of agents to original configuration to only attempt to restart changed agents instead of all of them. Some of them may not even be running.
• Add handling of what peer DID method is expected throughout the test and check for those in responses.
• Add tests that do not set the peer did on the responder and make sure it can respond to the peer DID of the requester.

AFJ Backchannel Updates

• Implement DID Exchange test harness API routes in the backchannel Increasing the tests run by AFJ #696
• Run DID Exchange tests with AFJ and add to the runsets Increasing the tests run by AFJ #696
• Implement Out of Band test harness routes in the backchannel. Increasing the tests run by AFJ #696
• Run Out of Band tests with AFJ and add to the runsets Increasing the tests run by AFJ #696
• Update the AFJ Backchannel based in the new AnonCreds API Update Credo AATH backchannel based on new AnonCreds API #635
• Update the restart functions to take any args passed through to start the agent with, including any peer:did setting.
  • Research is necessary to determine how AFJ sets the emitted peer did. It may not be at the agent startup, therefore we wouldn't need to restart the agent. However other equivalent work would need to be done where ever AFJ is told what peer did method to use.
• Create runsets for AFJ and ACA-Py to fully exercise qualified Peer DIDs.

VCX Backchannel Updates

• Apply the modifications above to the VCX Backchannel
  • Research is needed to determine what is exactly required.
• Create runsets for AFJ, ACA-Py, and VCX to fully exercise qualified Peer DIDs between these agents.

Findy Backchannel Updates?

• Apply the modifications above to the findy Backchannel?

AFGO Backchannel Updates

• Apply the modifications above to the AFGO Backchannel

[nodlesh]

This PR is ready for review and merge. We may have some help on the AFJ side so it would be good if these enhancements were in main.

These Qualified DID Tests will run in the regular interop pipeline test execution cycle for ACA-py. It will run in the full acapy runset. It will not run for any other runset until the backchannels have be brought up to scope for these tests.

Though these tests work with ACA-Py, we expect some of the test specification like the start parameters, etc to change to be more generalized across frameworks once we start working with AFJ and other frameworks in this test context.

[dbluhm]

LGTM!

[dbluhm]

Actually, comments: I'm not sure it makes sense for us to include tests for did:peer:3. Also, ACA-Py has not implemented an emit-did-peer-1 mode.

[nodlesh]

I've added them in for when they are implemented these tests should then automatically pass. However, if we prefer, we can just comment them out until or if we support those modes.
I like a test first approach, where we have failing tests then the software catches up to the tests to make them pass. I can understand that we may not want that here, showing a higher fail rate for unimplemented features.

[nodlesh]

@swcurran Would you prefer currently unsupported DID Peer Methods to be commented out or leave them failing? I'm inclined to think you would want them commented out.

[dbluhm]

We are unlikely to implement support for emitting a did:peer:1 or 3

[nodlesh]

Will any other Aries Framework implement those? The RFC mentions them so that is why I put them in.

[dbluhm]

AFJ implements 1 but it's actually not possible to exchange just did:peer:3s. You emit 2 and then derive a 3.

[nodlesh]

The last commit here comments out the DID Peer Methods that ACA-Py doesn't support. Will add them back if and when we support them or another Aries framework does.
Also loosened the check for the DID Peer Method to pass of fail a test scenario. The responders DID is just checked to make sure it is qualified, not that it has the same DID Peer Method as the requester.

[nodlesh]

@dbluhm or @swcurran Could I get a quick final approval of this? Would like to merge at this point. Thanks.

[swcurran]

Awesome work! Nice!