Retry logic and rate limit handling for all ATS APIs

[prPMDev]

ATS APIs can return transient errors (529, 503, 429). Currently no retry logic exists.

What to build

• Retry with exponential backoff (3 attempts, 1s/2s/4s delays)
• Handle 429 (rate limited): respect Retry-After header if present
• Handle 529/503 (server overload): retry after delay
• Handle timeouts: 10s per request, retry on timeout
• Log warnings on retries so users know what happened

Per-platform notes

• Greenhouse: no documented rate limits for public board API, but transient 529s observed
• Ashby: GraphQL endpoint, no known rate limits
• Lever: no known rate limits for public postings API

Why this matters

• For MCP server: AI agents may make rapid sequential requests
• For bulk registry verification: checking 100+ companies hits APIs fast
• For Pursuit integration: watchlist of 10+ companies fetches in parallel

Implementation

Add retry wrapper in a shared utility, used by all adapters:

async function fetchWithRetry(url, options, maxRetries = 3) {
  // exponential backoff, respect 429 Retry-After
}