v2: Program research, Groq extraction, and workout tracking (REVIEW — not deployed)

Backend:
- New models: ProgramCatalog, CatalogWorkout, CrawlQueue, WorkoutLog
- Program.py: added catalog_id, current_week, current_day columns
- program_research.py: ttp-crawler integration, Groq Llama 3.3 70B extraction, known sources map, validation
- crawl_scheduler.py: priority queue with off-peak/weekend scheduling
- catalog.py router: research, catalog browse, adopt, schedule, complete, progress endpoints
- Points: 75pts/workout, 50pts weekly bonus, 200pts completion bonus
- config.py: added groq_api_key setting
- main.py: catalog router, background scheduler, v2 migration

Frontend:
- ProgramSearch.jsx: search, research submission, catalog browsing, adopt programs
- ProgramDetail.jsx: schedule view, today's workout, completion modal, progress tracking
- App.jsx: Programs nav tab, new routes
- api.js: catalog and tracking API functions

Config:
- docker-compose.yml: GROQ_API_KEY env var

NOT DEPLOYED — requires Groq API key and Scot review

backend/app/config.py

@@ -21,6 +21,9 @@ class Settings(BaseSettings):
     polar_client_id: str = ""
     polar_client_secret: str = ""
 
+    # Groq (program extraction)
+    groq_api_key: str = ""
+
     # App
     base_url: str = "https://fitness.littlefake.com"
     timezone: str = "Asia/Bangkok"
@@ -31,3 +34,7 @@ class Settings(BaseSettings):
 @lru_cache
 def get_settings() -> Settings:
     return Settings()
+
+
+# Module-level shortcut used by services
+settings = get_settings()