Upgrade Chromium to 155.0.8059.79

[tnsardesai]

Summary

Bumps Chrome for Testing (and the matching ChromeDriver) from 152.0.7977.42 to 155.0.8059.79, the current Stable, in both the headful and headless images. The WebMCP fixture README named the version its behaviour was verified on, so it now says 155.

Stacked on #458. That PR fixes two Chromium 154 changes this bump would otherwise break: the headless PATCH /display viewport override being reset when its CDP session detaches, and the chrome://extensions Developer mode toggle changing from role=button to role=switch. Merge #458 first; this PR's base will then move to main.

Runtime dependencies

The only change to deb.deps between the two releases is the libglib2.0-0 minimum, from >= 2.39.4 to >= 2.49.3. Ubuntu 22.04 ships 2.72, so neither image needs new packages.

Validation

• CI: the server test workflow doesn't trigger on PRs targeting a non-main base, so I started it manually on this branch (run 38503931861, head f3cf83d, rebased onto Resize the headless viewport with Browser.setContentsSize #458's fallback fix). Both image builds, test-server-unit and test-server-e2e passed, including headless_default, headless_start_maximized, headless_kiosk and TestMV3ServiceWorkerRegistration.
• Local: built the headless image from this branch (chromium --version reports Google Chrome for Testing 155.0.8059.79) and ran the headless e2e tests against it with -parallel=4. 29 passed (before the Resize the headless viewport with Browser.setContentsSize #458 fallback fix), including TestPlaywrightExecuteAPI (all WebMCP subtests), the BiDi and CDP proxy tests, TestChromiumConfigureMultipartPowerset, TestExtensionUploadAndActivation and TestMV3ServiceWorkerRegistration. TestDisplayResizeChromiumWindow/headless_default failed once in that run, during navigateBlank: page.goto('about:blank') timed out after 30 s, before any resize was sent. It passed on 3 of 3 runs by itself, with the renderer at 1920x1080 and then 1280x720. The same pre-resize timeout happened once on 152, so it isn't caused by this bump.
• After the Resize the headless viewport with Browser.setContentsSize #458 fallback fix, the headless resize scenarios and MV3 also pass locally on this 155 image.
• I didn't run the headful image locally; CI covers it.