ci: Parallelize integration and smoke test shards

[Dmitriy Vasyura (dmitrivMS)]

Summary

• run Agent Host E2E in conditional platform jobs beside the remaining Electron integration tests
• split Electron smoke coverage into complementary core and chat shards on Linux, macOS, and Windows
• move the macOS PAC and Kerberos proxy scenarios off the main Electron smoke critical path
• skip repeated prelaunch synchronization in already-prepared test jobs
• avoid compiling Copilot for Browser smoke jobs, which do not register Chat or Agents Window suites
• skip login-shell environment probing for deterministic Browser and Remote test servers
• use Mocha's Dot console reporter for Browser unit tests while preserving full JUnit results

Measured impact

Using #334766 and this PR's first Code OSS attempt as baselines:

• the initial test sharding moved Agent Host E2E and chat/proxy smoke work off the main Electron paths
• Browser test servers currently repeat prelaunch synchronization 17-18 times per job
• a focused Browser smoke run took 15.4s with prelaunch and 12.2s without it, projecting roughly 50-60s saved per Browser job
• the Browser unit reporter benchmark reduced output from 2,761 to 263 lines and from 177KB to 80KB for the same 2,700-test subset, with identical JUnit coverage
• Browser-only Copilot compilation removes another 5-8s of unnecessary work

The remaining major setup opportunity is dependency caching: Windows downloads its 1.2GB node_modules cache in about 10s but spends about 71s extracting 6.6GB / 223k files. extensions/copilot/node_modules alone accounts for 1.28GB and 74k files, so job-specific additive caches could remove another 20-30s from Browser jobs, but that needs coordinated cache seeding and is intentionally not mixed into this change.

Validation

• parsed the root and reusable workflows and asserted caller inputs, check names, artifact uniqueness, prelaunch settings, and Browser-only compile conditions
• focused Browser smoke run: passed in 12.2s
• focused Remote smoke run: passed in 15.7s
• Browser unit benchmark: 2,700 tests passed with the Dot and JUnit reporters
• compiled test/integration/browser
• compiled test/automation
• compiled vscode-test-resolver
• targeted ESLint for all changed JavaScript and TypeScript files
• node --check test/smoke/test/index.js
• git diff --check
• npm ci

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

Note

This error may be related to your runner configuration. You can now configure runners for Copilot code review separately from Copilot cloud agent by creating a copilot-code-review.yml file with your setup steps. Read the docs for details.

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

Note

This error may be related to your runner configuration. You can now configure runners for Copilot code review separately from Copilot cloud agent by creating a copilot-code-review.yml file with your setup steps. Read the docs for details.