Scrape a page that needs browser interaction first (click, scroll, wait, type) using Firecrawl scrape actions

Documented steps

1. AUTH: Authorization: Bearer fc-YOUR_API_KEY; base URL https://api.firecrawl.dev/v2
2. POST https://api.firecrawl.dev/v2/scrape with an 'actions' array executed in order before formats are captured
3. Example: {"url":"https://example.com","formats":["markdown"],"actions":[{"type":"click","selector":"#load-more"},{"type":"wait","milliseconds":2000},{"type":"scroll","direction":"down"}]}
4. To type into a field: {"type":"click","selector":"input#search"} then {"type":"write","text":"query"}, optionally {"type":"press","key":"Enter"} — 'write' requires a prior click to focus the field
5. Wait for elements with {"type":"wait","selector":"#el"} or for time with {"type":"wait","milliseconds":N}
6. Capture mid-sequence state with a {"type":"screenshot",...} or {"type":"scrape"} action; {"type":"executeJavascript","script":"..."} runs custom JS
7. Final page state lands in the requested top-level formats; per-action captures appear under data.actions (e.g. data.actions.screenshots[])