Run reports and audit

# Audit the most recent run
sverka audit

# Audit a specific run
sverka audit run-abc123

# JSON output for programmatic consumption
sverka audit --format json

Run: run-abc123
Plan: rp-ci-pipeline
Status: success
Duration: 45.2s

Steps:
  ci/lint      1.2s   cache: hit
  ci/test      8.4s   cache: miss
  ci/build    12.1s   cache: hit
  ai/review   18.3s   cache: - (non-cacheable)
                       tokens: 12,400 in / 3,200 out
                       model: gpt-4o
                       estimated cost: $0.063

Aggregate:
  Total duration: 45.2s
  Cache hit rate: 67% (2 of 3 cacheable steps; non-cacheable steps excluded)
  AI token usage: 15,600 tokens
  Estimated AI cost: $0.063

interface RunReport {
  readonly runId: string;
  readonly planId: string;
  readonly status: RunStatus;
  readonly startedAt: number;
  readonly completedAt: number;
  readonly durationMs: number;
  readonly steps: readonly StepReport[];
  readonly summary: RunSummary;
}

interface StepReport {
  readonly stepId: string;
  readonly status:
    "succeeded" | "failed" | "skipped" | "cancelled" | "cache-hit";
  readonly durationMs: number;
  readonly cacheHit?: boolean;
  readonly cacheKey?: string;
  readonly retries?: number;
  readonly agent?: AgentReport;
}

interface AgentReport {
  readonly engine: string;
  readonly model?: string;
  readonly inputTokens?: number;
  readonly outputTokens?: number;
  readonly totalTokens?: number;
  readonly estimatedCostUsd?: number;
}

interface RunSummary {
  readonly totalSteps: number;
  readonly succeeded: number;
  readonly failed: number;
  readonly skipped: number;
  readonly cacheHits: number;
  readonly cacheMisses: number;
  readonly totalDurationMs: number;
  readonly totalTokens?: number;
  readonly estimatedCostUsd?: number;
}