Fix M33MU console marker failures in CI

[aidangarske]

• Currently m33mu tests have been flaking because of mixed console outputs that get read wrong and report a CI failure
• Rejoin guest0 markers interrupted by UART banners or guest1 output.
• Share console parsing between CI and the local runner.
• Preserve raw logs and normal line boundaries.
• Reject missing bytes, faults, and missing guest1 markers.
• Pass 11 regression tests and 75 captured-log checks.
• Pass 200 hosted boots across both engines and guest variants under CPU load. (original run failed a few)
• Should actually close M33MU runs occasionally lose console output or shift interrupt timing on GitHub runners #34.

[Copilot]

Copilot review overview

🟡 Changes recommended

Several local guest0 and conformance checks still bypass reconstruction, and the changes do not address the linked issue’s documented runtime failure.

Review effort: Balanced
Findings: 1

Open (1)

• Feed confboot result parsing from the reconstructed guest0 log · New

What changed in this PR

Adds shared M33MU console reconstruction to reduce false marker failures caused by interleaved UART output.

Changes:

• Reconstructs guest0 output while preserving raw logs.
• Integrates parsing into local and CI runners.
• Adds regression tests and CI execution.

File	Description
tests/​target/​run_m33mu_scenario.sh	Uses reconstructed output for selected checks.
tests/​target/​lib/​m33mu_console.py	Implements console reconstruction.
tests/​scripts/​test_m33mu_console.py	Adds parser regression tests.
Makefile	Runs parser tests from make test.
.github/​workflows/​unit-tests.yml	Adds parser test coverage in CI.
.github/​workflows/​m33mu.yml	Uses reconstructed output for CI markers.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.