Improve CLAUDE.md with richer project context

Problem

The current CLAUDE.md (169 lines) provides basic project structure and commands but is missing context that would make AI agents significantly more effective when working on this codebase.

Suggested Improvements

• Key architecture decisions — why certain patterns were chosen (e.g., domain-based org isolation, X-RH-IDENTITY flow, content guard design)
• Common pitfalls — things that trip up new contributors (e.g., migration ordering, test database setup, fixture requirements)
• Code conventions — naming patterns, serializer/viewset patterns used consistently across the codebase
• Testing guidance — how to run specific test subsets, what fixtures are available, how to set up test data
• Deployment context — how the three-service architecture (api/content/worker) affects development decisions
• Dependency relationships — how pulp-service relates to pulpcore, what's inherited vs. overridden
• Recent major changes — any in-progress migrations or deprecations that agents should be aware of

The agent should read the existing CLAUDE.md, explore the codebase thoroughly (especially docs/ARCHITECTURE.md), and produce an improved version that gives AI agents the context they need to make good decisions.

Read the existing CLAUDE.md and docs/ARCHITECTURE.md first, then inspect the codebase to verify the architecture, conventions, testing setup, deployment services, dependency relationships, and recent changes listed in the issue. Done means CLAUDE.md is expanded with accurate, actionable context for AI agents without omitting the requested areas.