improve copilot agent setup: instructions, prompts, and ci environment

[hanniavalera]

This pull request adds comprehensive, step-by-step contributor documentation for two common extension tasks: adding a new command and adding a new diagnostic/output parser. It also significantly expands and clarifies the project conventions and architecture guide in .github/copilot-instructions.md, covering configuration handling, kit trust, localization, diagnostics, and testing.

The most important changes are:

Contributor Documentation Additions:

• Added .github/prompts/add-command.prompt.md, a detailed guide for adding a new cmake.* command, including required file changes, code patterns, conventions, and a verification checklist.
• Added .github/prompts/add-diagnostic-parser.prompt.md, a step-by-step recipe for implementing and registering a new diagnostic/output parser, with code examples, registration steps, testing, and a checklist.

Project Conventions and Architecture Guide Improvements in .github/copilot-instructions.md:

• Expanded configuration and error handling conventions: clarified path aliases, error reporting patterns (rollbar usage), and introduced a new section on reactive configuration subscriptions using ConfigurationReader.onChange() instead of direct VS Code events. [1] [2]
• Added documentation on handling sentinel kit values (SpecialKits), the kit trust model, and generator types, including the use of util.isMultiConfGeneratorFast() and respecting the cmake.setBuildTypeOnMultiConfig setting.
• Clarified localization (NLS) conventions, including NLS key naming, updating package.nls.json, and rules for markdownDescription.
• Added guidelines for setting scope in package.json for configuration settings, instructions for adding a new diagnostic parser, and rules for handling Copilot CI environment artifacts.
• Expanded the testing checklist with a new section on backend tests (yarn backendTests), including file locations, import strategies, and when to mirror logic for VS Code dependencies.