test: add coverage for Python/Node runtimes and upload-pipeline-artifact

[Copilot]

Closes test gaps in the Python/Node runtimes and upload-pipeline-artifact: zero unit tests on the YAML generators, no integration coverage for runtimes: python|node, no tools/list assertion for upload-pipeline-artifact, and no execute dispatch test for it.

• Python/Node generator unit tests — generate_python_install (default + pinned), generate_pip_authenticate, generate_node_install (default + pinned), generate_npm_authenticate, generate_ensure_npmrc (default registry fallback + custom feed-url branch).
• Compile integration tests — runtimes: python: true, runtimes: python: { version: "3.12" }, and runtimes: node: true end-to-end via the ado-aw binary, asserting the expected UsePythonVersion@0 / NodeTool@0 task and versionSpec.
• MCP tools/list — assert upload-pipeline-artifact is registered.
• Execute dispatch — test_execute_malformed_upload_pipeline_artifact_returns_err mirrors the existing wiki/upload patterns and exercises the dispatch_resource_tools arm.
• Bug fix surfaced by the new Node integration test — NodeExtension::name() returned "Node.js", whose . is rejected by the wrap_prompt_append display_name sanitizer, so any pipeline with runtimes: node failed to compile. Renamed to "Node" and updated the three pre-existing extension tests pinning the string. The Node runtime had no compile-level test, which is why this slipped through — exactly the gap this PR closes.
• Merge conflict resolution — merged main into the branch and resolved a conflict in tests/compiler_tests.rs, preserving both the new dotnet runtime integration tests landed on main and the new Python/Node runtime integration tests added by this PR.

Summary

Adds 7 unit tests, 3 compile-level integration tests, and a missing MCP tools/list assertion to cover the Python/Node runtimes and the upload-pipeline-artifact safe output. The new Node integration test surfaced and is fixed alongside a regression where NodeExtension::name() ("Node.js") violated the display_name sanitizer, breaking Node runtime compilation. Also resolves a merge conflict with main so both the new dotnet runtime tests and the new Python/Node runtime tests coexist in tests/compiler_tests.rs.

Test plan

• cargo test --bin ado-aw — 1234 passed (was 1227, +7).
• cargo test --test compiler_tests -- test_python_runtime test_node_runtime test_dotnet_runtime test_lean_runtime — all 6 runtime integration tests pass after the merge resolution.
• cargo test --test compiler_tests -- test_python_runtime test_node_runtime — 3 new tests pass; without the Node.js → Node fix, test_node_runtime_compiled_output fails with the sanitizer error, confirming the test catches the regression.
• cargo test --test mcp_http_tests test_mcp_initialize_and_tools_list — passes with the new assertion.
• cargo clippy --all-targets --all-features — no new warnings.

[jamesadevine]

@copilot resolve the merge conflicts in this pull request

[Copilot]

@copilot resolve the merge conflicts in this pull request

Resolved the merge conflicts in tests/compiler_tests.rs by preserving both the new dotnet runtime integration tests from main and the new Python/Node runtime integration tests from this branch. All six runtime integration tests pass (test_dotnet_runtime_compiled_output, test_dotnet_runtime_with_feed_url_compiled_output, test_python_runtime_compiled_output, test_python_runtime_pinned_version_compiled_output, test_node_runtime_compiled_output, test_lean_runtime_compiled_output). Done in 16bd337.