Eradicate mediated connection from issuance and presentation integration tests, part 1

[mirgee]

Reduces the number of issuance and presentation integration tests relying on mediated connection to exchange messages instead of exchanging them directly between the SM handlers, with the goal of removing their dependency on agency client and MediatedConnection and eventually making the integration tests runnable without a mediator.

This change results in a minor test speedup, however the main bottleneck is schema / cred. def. creation, which can be mediated by sharing schemas / credential definitions between tests wherever possible.

[codecov-commenter]

Codecov Report

Merging #967 (9cb8cd5) into main (c08650a) will decrease coverage by 4.19%.
Report is 2 commits behind head on main.
The diff coverage is 71.61%.

@@            Coverage Diff             @@
##             main     #967      +/-   ##
==========================================
- Coverage   39.32%   35.13%   -4.19%     
==========================================
  Files         414      413       -1     
  Lines       28863    27471    -1392     
  Branches     6183     5811     -372     
==========================================
- Hits        11349     9653    -1696     
- Misses      14175    14776     +601     
+ Partials     3339     3042     -297     

Flag	Coverage Δ
unittests-aries-vcx	35.13% <71.61%> (-4.19%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files Changed	Coverage Δ
aries_vcx/src/common/test_utils.rs	94.22% <ø> (+0.67%)	⬆️
aries_vcx/src/handlers/issuance/holder.rs	42.67% <ø> (-17.33%)	⬇️
aries_vcx/src/handlers/out_of_band/receiver.rs	25.16% <0.00%> (ø)
aries_vcx/src/handlers/util.rs	12.94% <ø> (-4.71%)	⬇️
...ation/verifier/states/presentation_request_sent.rs	55.26% <0.00%> (ø)
aries_vcx/src/utils/devsetup.rs	43.35% <ø> (-4.20%)	⬇️
aries_vcx/tests/test_agency.rs	77.60% <ø> (ø)
aries_vcx/tests/test_pool.rs	83.82% <ø> (ø)
aries_vcx/tests/utils/devsetup_alice.rs	86.44% <ø> (+13.80%)	⬆️
aries_vcx/tests/utils/devsetup_faber.rs	85.29% <ø> (+4.62%)	⬆️
... and 7 more

... and 54 files with indirect coverage changes

[Patrik-Stas]

this is amazing 🚀 Significant improvement in readability and mental overhead reading the tests, now that connections are gone.

Left some comments but generally ✅

[Patrik-Stas]

Goddamn Jimmy, this is some serious gourmet shiiiit!

I am even looking at integration-tests-cntd branch which seems to be trimmed down even a lot more than this. I might be getting ahead of myself, because I don't know what's your working strategy, but noticed on that branch, mysql tests are also removed - I'd keep that; do you want to remove them?

One more intermediate remark, and again maybe you already plan doing something like this, just sharing my current thoughts - relating to our private discussion yesterday about scenario.rs:

• I would suggest to move methods which related to a single agent as an Agent's impl method (and ideally merge Alice & Faber into one singular Agent abstraction); such as

pub async fn accept_proof_proposal(
    faber: &mut Faber,
    verifier: &mut Verifier,
    presentation_proposal: AriesMessage,
) -> AriesMessage {

• While keeping in workflows methods which contain 2-agent interaction, such as

pub async fn _exchange_credential(
    consumer: &mut Alice,
    institution: &mut Faber,
    credential_data: String,
    cred_def: &CredentialDef,
    rev_reg: &RevocationRegistry,
    comment: Option<&str>,
) -> Issuer {

[mirgee]

I am even looking at integration-tests-cntd branch which seems to be trimmed down even a lot more than this. I might be getting ahead of myself, because I don't know what's your working strategy, but noticed on that branch, mysql tests are also removed - I'd keep that; do you want to remove them?

No worries, mysql tests will be preserved to the extent they are useful, as discussed :)

• I would suggest to move methods which related to a single agent as an Agent's impl method (and ideally merge Alice & Faber into one singular Agent abstraction); such as ...

I totally agree that this is the desired final state, pruning the current Alice and Faber just an intermediate step. They will next be merged into a unified test agent struct, followed by implementation of single-agent-methods and refactoring of currently so-called "scenarios".

Another (probably less favorable) option would be to get rid of test agent representation altogether for now (perhaps keeping it just as a data structure holding profile equivalents and whatever handlers necessary for the test at hand), awaiting a "more final", generic agent implementation (i.e. useful outside of tests as well).