Blueprint for chaos

[gloskull]

Motivation

• Provide a code-backed, single source of truth for staging chaos experiments so off-chain orchestrators and runbooks share the same safety thresholds.
• Define system-wide safety gates (latency, availability, error-budget, healthy zones, security approvals) and guarded blue-green/canary rollout phases for safe staging experiments.
• Document the staging experiment scope and operational runbook so runs can be audited and automatically gated.

Description

• Add a new module src/chaos.rs that defines constants, ServiceSurface, FaultKind, RolloutPhase, ChaosExperiment, ChaosHealthSnapshot, a canonical STAGING_CHAOS_EXPERIMENTS catalog, and next_rollout_phase and safety-gate helpers.
• Register the new module from the crate root by adding pub mod chaos; in src/lib.rs so the catalog and helpers are available to off-chain tooling.
• Add human-facing documentation at docs/chaos-engineering-staging.md describing architecture, scope, safety gates, monitoring/alerting, blue-green/canary flow, and a runbook.
• Include unit tests in src/chaos.rs that validate catalog coverage, safety-gate behavior, and rollout progression.

Testing

• Ran cargo test chaos which executed the chaos module unit tests catalog_covers_every_service_surface_with_safe_blast_radius, safety_gate_enforces_latency_availability_security_and_canary_burn, and rollout_phase_advances_through_blue_green_and_canary, and all passed.
• The run also executed the existing mempool chaos test chaos_50k_transactions_block_matches_top_fee_per_gas_subject_to_gas_limit and it passed.
• Overall the automated test run completed with the chaos-related tests passing and no test failures.

Closes #80