This commit introduces a new `WebScraperService` to enable programmatic interaction with web content loaded within a `<browser>` element.

The service is built using a JSWindowActor pair (`NRWebScraperParent` and `NRWebScraperChild`) for robust communication between the parent process and the content process. This allows for automating tasks like data extraction and form filling.

The initial implementation includes methods to:
- Retrieve the full HTML of a page.
- Input values into elements using CSS selectors.

Supporting changes include:
- Integrating the service via `BrowserGlue`.
- Expanding `XULBrowserElement` TypeScript definitions to improve type safety for browser interactions.
- Updating the Solid XUL runtime to allow the `messagemanagergroup` attribute required for actors.

This commit introduces a comprehensive web scraping service to enable browser automation and data extraction.

The implementation is split into two main components:
- `WebScraperService`: A main process module that manages browser instances and exposes a high-level API for scraping operations.
- `NRWebScraperChild`: A JSWindowActor that runs in the content process to safely execute DOM manipulations and data extraction tasks.

This new service provides the following capabilities:
- Get full page HTML or text from specific elements.
- Interact with elements (click, input text).
- Wait for an element to appear before proceeding.
- Execute custom JavaScript in the page context.
- Take various types of screenshots (viewport, element, full-page, region).

Supporting Gecko type definitions for `CSSStyleDeclaration` and `XULBrowserElement` have also been updated.

This commit introduces a new OS API layer designed to provide browser context, likely for features like LLM-generated workflows.

A central `OSGlue` module is added to manage the initialization of various OS-level services. The first major service implemented is `BrowserInfo`, which can collect recent tabs, browsing history, and downloads.

This new system is initialized during application startup. As part of this change, the existing `WebScraperService` is refactored into the new `os-apis` structure.

Rename the `WebScraperService.sys.mts` file to `WebScraperServices.sys.mts` to better reflect that the module manages multiple scraper instances.

The import path in `OSGlue.sys.mts` has been updated to match the new filename.

Implement a right-click context menu for items in the Top Sites grid. This provides users with quick access to actions like pinning/unpinning sites, opening links in new tabs/private windows, and blocking a site from appearing.

A new "Blocked Sites" section has been added to the Settings modal, allowing users to view and unblock sites.

This also refactors data management for the new tab page. Settings are now centralized into a single object managed by `getNewTabSettings` and `saveNewTabSettings` in `utils/dataManager.ts`, simplifying state persistence. The previous `TopSites/dataManager.ts` has been removed.

This commit updates the restart functions in the RebootPanelMenu to use modern, recommended APIs, improving robustness and aligning with current platform practices.

- Replace direct environment variable manipulation for cache clearing with the `Services.appinfo.invalidateCachesOnRestart()` method.
- Use the `restart-in-safe-mode` observer notification instead of the direct `Services.startup.restartInSafeMode()` call.
- Use the `eAttemptQuit` flag instead of `eForceQuit` to allow for a more graceful application shutdown.
- Remove the unused "Restart with Profile Manager" option and its corresponding logic.