Add backtrace capturing, prototype a test err handling

[Patrik-Stas]

• Removed run(fn) test case wraps as they turned out to be no longer necessary. Deleted related dead code.
• Add feature backtrace_errors; when enabled, backtrace dependency is included and error thrown from aries-vcx contain capture backtrace information
• Tweak Display impl for AriesVcxError
• Add Debug impl for AriesVcxError - this is useful when err is thrown from a test run by cargo test runner
• Modified all integration tests:
  • They return -> Result<(), Box<dyn Error>> instead of -> ().
  • Therefore we can write test code without being coerced into calling unwrap()
  • The only good thing about unwrap() panics was that that they included stacktrace, which is not default with returning an error - but with additions mentioned above, we are not loosing this capability

If agreed, we can continue adopting:

• same panic-less approach in the rest of the test code
• backtrace capturing for other sub-crates as well

After rebase

I've temporarily added run2 but I want to question 2 things:

• What's the purpose of macro run_setup_test(...) as opposed to just calling build_setup_profile().await.run(...)
• I believe that whole run thing pub async fn run<F>(self, f: impl FnOnce(Self) -> F) is now obsolete, there's no universal setup&cleanup which is valid for every test. In fact, run is not doing anything useful atm, other the running the test itself

[codecov-commenter]

Codecov Report

Merging #1022 (2939007) into main (4e2a5f4) will increase coverage by 0.00%.
The diff coverage is 0.36%.

❗ Current head 2939007 differs from pull request most recent head cbe30ca. Consider uploading reports for the commit cbe30ca to get more accurate results

@@          Coverage Diff           @@
##            main   #1022    +/-   ##
======================================
  Coverage   0.05%   0.05%            
======================================
  Files        384     384            
  Lines      21249   21084   -165     
  Branches    3821    3910    +89     
======================================
  Hits          12      12            
+ Misses     21236   21071   -165     
  Partials       1       1            

Flag	Coverage Δ
unittests-aries-vcx	0.05% <0.36%> (+<0.01%)	⬆️

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

Files	Coverage Δ
aries_vcx/src/handlers/mediated_connection/mod.rs	0.00% <ø> (ø)
...vcx/src/protocols/issuance/issuer/state_machine.rs	0.00% <ø> (ø)
tools/test_utils/src/devsetup.rs	0.00% <0.00%> (ø)
aries_vcx/tests/test_mysql_wallet.rs	4.00% <0.00%> (+0.15%)	⬆️
aries_vcx/src/errors/error.rs	0.00% <0.00%> (ø)
...es_vcx/src/common/proofs/prover/prover_internal.rs	0.00% <0.00%> (ø)
aries_vcx/tests/test_connection.rs	0.87% <1.66%> (ø)
aries_vcx/tests/test_primitives.rs	1.28% <1.42%> (+0.10%)	⬆️
aries_vcx/tests/test_credentials.rs	1.02% <1.12%> (+0.08%)	⬆️
aries_vcx/tests/test_anoncreds.rs	0.89% <1.01%> (+0.06%)	⬆️
... and 6 more

[bobozaur]

@Patrik-Stas Please note that as part of #1023 I actually copied the test_setup() function from here. I initially wanted to pull some trickery and allow Arc<dyn Trait> in tests but I forgot that the generic arguments in the traits methods make them not trait object safe, so we have to stick with generics. Since I'm moving some things around I figured it could be easier if I copied the implementation there.

[bobozaur]

* What's the purpose of macro `run_setup_test(...)` as opposed to just calling `build_setup_profile().await.run(...)`

None, really. Just convenience.

* I believe that whole run thing `pub async fn run<F>(self, f: impl FnOnce(Self) -> F)` is now obsolete, there's no universal setup&cleanup which is valid for every test. In fact, `run` is not doing anything useful atm, other the running the test itself

I assume it could be removed in this case, indeed.

[Patrik-Stas]

This should be ready for re-review.

• Removed the setup run(fn) wraps as they turned out unnecessary
• Removed unwrap() from all tests in favor of returning error

[gmulhearn]

Lgtm