test: expand workflow runner coverage

[man4ish]

Summary

• Expanded meaningful unit coverage for the workflow runner with 32 new tests.
• Updated coverage measurement to include both omni_tool_runtime and tools.
• External execution boundaries are mocked; no Docker, Apptainer/Singularity, cloud, HPC, or network infrastructure is required.
• No production source files were modified.
• No coverage exclusions were added.

Tests

• Focused workflow-runner tests: 32 passed.
• Full suite: 403 passed, 0 failed, 0 skipped.
• Full suite collected 403 tests.
• Final coverage: 98.60% statements/lines, 95.15% branches, 96.55% functions.

Coverage

• Baseline: 371 tests and 99.56% existing measured coverage.
• Final: 98.60% statements/lines, 95.15% branches, 96.55% functions.
• The final measurement includes the previously unmeasured tools/workflow_runner/run.py production module.
• Branch coverage remains enabled.

Scope and remaining gaps

• Tests cover execution lifecycle, command construction, local/cloud staging, input/output handling, upload/download boundaries, environment propagation, result handling, and failure paths.
• Remaining gaps are narrow defensive branches and real external execution integrations, which remain integration-only by design.
• No confirmed production bugs were discovered.

Verification

• pytest --no-cov -q tests/test_workflow_runner.py: 32 passed.
• Full coverage suite: 403 passed, 0 failed, 0 skipped; 98.60% statements/lines, 95.15% branches, 96.55% functions.
• Ruff passes for tests/test_workflow_runner.py.
• Full-repository Ruff still reports pre-existing issues, including findings in unchanged production files; those were not modified in this PR.

Production protection

Production source modifications: 0.

This PR is test-only and is left open for review. It must not be merged as part of this task.