[FEAT]: Performance Tests

[SeedyROM]

Changes:

• Refactor kiko-backend to be a library and have an executable.
• Write some criterion benchmarks for the most critical systems.

[Copilot]

Pull Request Overview

This PR adds comprehensive performance testing infrastructure using the Criterion benchmarking library. The changes include refactoring the kiko-backend crate to be a library with an executable, and implementing benchmarks for critical systems including ID generation, session management, and PubSub messaging.

Key changes:

• Refactored kiko-backend to expose public library API while maintaining executable functionality
• Added Criterion benchmark suites for ID generation, session service operations, and PubSub messaging
• Updated documentation with performance testing guidance and instructions

Reviewed Changes

Copilot reviewed 9 out of 10 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
crates/kiko/benches/id_generation.rs	Comprehensive benchmarks for ID generation performance, uniqueness validation, and serialization
crates/kiko-backend/benches/session_service.rs	Performance tests for session CRUD operations, participant management, and concurrent access
crates/kiko-backend/benches/pubsub.rs	Benchmarks for PubSub messaging throughput, concurrent operations, and memory cleanup
crates/kiko-backend/src/lib.rs	New library entry point exposing public API
crates/kiko-backend/src/main.rs	Refactored executable to use library modules
crates/kiko-backend/src/services/sessions.rs	Added Clone derive for benchmarking compatibility
crates/kiko/Cargo.toml	Added Criterion dependencies and benchmark configuration
crates/kiko-backend/Cargo.toml	Added dev dependencies for async benchmarking
CLAUDE.md	Updated documentation with performance testing commands and guidance

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[Copilot]

Converting to string via id.as_str().to_string() is inefficient for uniqueness testing. Consider inserting the ID directly into HashSet<SessionId> since IDs likely implement Hash and Eq, which would avoid unnecessary string allocations.

[Copilot]

Converting to string via id.as_str().to_string() is inefficient for uniqueness testing. Consider inserting the ID directly into HashSet<ParticipantId> since IDs likely implement Hash and Eq, which would avoid unnecessary string allocations.

[Copilot]

Cloning the string for each iteration is inefficient in a benchmark. Consider using SessionId::from_string(s) or SessionId::from(s.as_str()) to avoid unnecessary allocations during performance measurement.

Suggested change

	black_box(SessionId::from_string(s.clone()));
	black_box(SessionId::from_string(s));