[BUG] OWL Test Harness docker orchestration issue

[PatStLouis]

The ACA-py BDD Interop tests started to consistently fail a couple of days ago.
https://github.com/openwallet-foundation/acapy/actions/runs/22083629309/job/63813652859?pr=4050

[PatStLouis]

I think its fixed now, waiting on #4052 to complete

BDD Interop Integration Tests failing due to Universal Resolver crash on Ubuntu 24.04 runners

GitHub Actions runners updated to Ubuntu 24.04 (Feb 8, 2026), which uses cgroups v2 by default. The Universal Resolver's uni-resolver-web Docker image (0.8.0-93c2f32, from August 2022) bundles a JDK that lacks cgroups v2 support, causing a NullPointerException at cgroupv2.CgroupV2Subsystem.getInstance() on startup. The container exits immediately with code 1, the manage script times out waiting for the resolver health check, and the entire test run fails before any BDD tests execute.

Additionally, the manage script had several bash issues that made failures harder to diagnose:

• Missing $ on variable references (IS_*_REMOTE, DATA_VOLUME_PATH) causing conditions to always evaluate as true
• docker stop/rm commands failing the script when containers don't exist (no || true)
• Incorrect error suppression (|| 1 instead of || true)
• No diagnostic output when the Universal Resolver fails to start

Fix (owl-agent-test-harness)

• Pin uni-resolver-web to 1.1.0-86d4765 and driver-did-sov to 0.15.0-9e56cd9 (cgroups v2-compatible JDKs)
• Fix all missing $ on variable references in the manage script
• Add || true to Docker cleanup commands to prevent premature script exit
• Add diagnostic logging (container status + logs) when the Universal Resolver fails to start

Fix (acapy)

• Add LEDGER_TIMEOUT=180 to give the Universal Resolver more startup time on CI
• Stream test output with 2>&1 | tee output.txt for better CI log visibility
• Run results check with if: always() to ensure failures are always reported
• Add output.txt existence check to detect tests that didn't run