Audit: Add text/plain content negotiation

[nycomp]

Audit: Add text/plain Content Negotiation

Parent: #424 (Audit System)
Related: #427 (Routes - JSON only)

Overview

Add text/plain response format support to all audit GET endpoints for CLI-friendly output. Currently only application/json is supported.

Tasks

1. Content Negotiation Helper

• Create Accept header parser utility
• Handle Accept: application/json, text/plain, */*
• Default to JSON for unspecified/invalid Accept headers

2. Text Response Formatters

GET /traces - Compact trace list:

trace abc123  142.5ms  OK   client:timetable-app  user:js
trace def456   15.8ms  FAIL client:—              user:—

--- more: cursor=2026-03-27T09:58:12.000Z ---

GET /traces/ - Waterfall format:

trace abc123  142.5ms  OK  client:timetable-app  user:js

+0.0ms    142.5ms  GET /api/v1/students           200
+0.1ms    15.2ms     POST /auth/token/validate     200
+18.0ms   98.3ms     GET /api/v1/db/query          200

GET /traces/search - Same as GET /traces

GET /health - Simple:

OK

Acceptance Criteria

• Accept: application/json returns JSON (current behavior)
• Accept: text/plain returns formatted text
• Invalid/missing Accept header defaults to JSON
• Content-Type header matches response format
• Contract tests cover both formats

Implementation Notes

• Text formatters should accept model objects (not resource dicts)
• Reuse TraceSummary, TraceTree, TraceSpan models
• Format strings using f-strings for maintainability
• Status: "OK" (200-299), "FAIL" (500+), "ERR" (400-499)

References

• PRD §5.3: Content Negotiation
• PRD §6: Trace Reconstruction (waterfall format)