feat: AI advisor integration and architecture improvements

[yairfalse]

Summary

• AI advisor integration: Progressive 4-level adoption ladder (Off → Context → Advised → Planned → Driven) for AI-assisted rollout decisions. Level 0 preserves current behavior. Level 1 enriches FALSE Protocol events. Level 2 consults an external advisory service while thresholds still decide.
• A/B testing strategy: Full A/B testing support with header/cookie-based routing, statistical significance evaluation, and experiment lifecycle management
• FALSE Protocol occurrences: Rich AI-native events with step/weight/traffic context, possible causes, and suggested fixes
• Architecture cleanup: Extracted reconcile loop into focused modules (replicaset, status, traffic, validation), added clock abstraction, Prometheus A/B metrics
• Documentation: Rewritten CLAUDE.md and README to match current architecture

Test plan

• 301 unit tests pass (cargo test)
• Zero compiler warnings
• CRD schema regenerated and test passes
• cargo fmt clean
• Manual: apply a Rollout with advisor.level: Context and verify enriched occurrences
• Manual: apply a Rollout with advisor.level: Advised and verify advisor consultation logs

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR refactors the FALSE Protocol occurrence handling in KULTA by replacing ~140 lines of hand-rolled structs with the canonical false-protocol crate, bringing spec compliance and a single source of truth. The changes also include documentation updates to CLAUDE.md and minor code formatting improvements across test files.

Changes:

• Replaced inline FALSE Protocol type definitions with false-protocol crate dependency
• Refactored build_occurrence() to use the crate's builder pattern and constructors
• Updated CLAUDE.md to be more concise and reflect current architecture
• Applied cargo fmt formatting across test files (rollout_test.rs, prometheus_ab.rs, cdevents_test.rs)

Reviewed changes

Copilot reviewed 6 out of 7 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
Cargo.toml	Added false-protocol crate as a path dependency, removed ulid direct dependency
Cargo.lock	Added false-protocol package entry with its transitive dependencies
src/controller/occurrence.rs	Major refactor: removed ~110 lines of struct definitions, updated to use false-protocol types and builder methods
src/controller/rollout_test.rs	Formatting improvements: reformatted long function calls to fit line length conventions
src/controller/prometheus_ab.rs	Formatting improvements: reformatted function calls for readability
src/controller/cdevents_test.rs	Formatting improvements: reformatted assertion chains
CLAUDE.md	Rewrote from verbose learning guide to concise reference documentation

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The false-protocol dependency uses a relative path (../false-protocol/rust), which makes the project unbuildable unless the false-protocol crate is available at that specific relative location. This creates tight coupling and makes it difficult for others to build, test, or contribute to this project.

Consider one of these alternatives:

1. Publish the false-protocol crate to crates.io and reference it by version
2. Use a git dependency with a repository URL if the crate is in a separate repo
3. Move the false-protocol code into a workspace if both crates are meant to be maintained together
4. Document the repository structure requirements clearly in the README if the relative path is intentional for development