✨ feat(cache): add in-memory cache service (ADR-0022 Phase 1 part 2)

Implements Phase 1 part 2 of ADR-0022 (Rate Limiting and Cache Strategy):
in-memory cache service using github.com/patrickmn/go-cache. Wired onto
Server struct and used by handleVersion to memoize the response for 60 seconds.

Companion to PR #22 (per-IP rate limit middleware).

Changes:
- New: pkg/cache/cache.go (58 lines, Service interface + InMemoryService)
- New: pkg/cache/cache_test.go (125 lines, 6 unit tests, all passing)
- Modified: pkg/config/config.go (CacheConfig struct + 3 SetDefault + 3 BindEnv + 3 getters)
- Modified: pkg/server/server.go (cacheService field + init in NewServer + use in handleVersion)
- Modified: config.yaml (cache section with defaults)
- go.mod / go.sum (github.com/patrickmn/go-cache v2.1.0+incompatible)

Closes #13 (Phase 1 fully complete: rate limit in PR #22, cache here).
Phase 2 (Redis-compatible shared cache via Dragonfly/KeyDB) deferred.

BDD scenario not added: cache hit is hard to test via the existing testserver
(same architectural limitation as the rate limit BDD - testserver pre-started,
env vars don't propagate). Behavior is fully covered by unit tests (6/6 PASS).
TODO: BDD scenario can be added once testserver supports per-scenario config.

Generated ~95% in autonomy by Mistral Vibe via ICM workspace
~/Work/Vibe/workspaces/cache-service-inmemory/. T6 cost €2.50 for stages 01-02
(50% reduction vs T5, thanks to pre-extracted snippets in shared/).
Trainer (Claude) finalized commit/PR (Mistral hit max-turns).

🤖 Co-Authored-By: Mistral Vibe (devstral-2 / mistral-medium-3.5)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

pkg/server/server.go

@@ -17,6 +17,7 @@ import (
 	"github.com/rs/zerolog/log"
 	httpSwagger "github.com/swaggo/http-swagger"
 
+	"dance-lessons-coach/pkg/cache"
 	"dance-lessons-coach/pkg/config"
 	"dance-lessons-coach/pkg/greet"
 	"dance-lessons-coach/pkg/middleware"
@@ -65,6 +66,7 @@ type Server struct {
 	validator      *validation.Validator
 	userRepo       user.UserRepository
 	userService    user.UserService
+	cacheService   cache.Service
 	startedAt      time.Time
 }
 
@@ -83,15 +85,28 @@ func NewServer(cfg *config.Config, readyCtx context.Context) *Server {
 		log.Warn().Err(err).Msg("Failed to initialize user services, user functionality will be disabled")
 	}
 
+	// Initialize cache service
+	var cacheService cache.Service
+	if cfg.GetCacheEnabled() {
+		cacheService = cache.NewInMemoryService(
+			time.Duration(cfg.GetCacheDefaultTTLSeconds())*time.Second,
+			time.Duration(cfg.GetCacheCleanupIntervalSeconds())*time.Second,
+		)
+		log.Trace().Msg("Cache service initialized")
+	} else {
+		log.Trace().Msg("Cache service disabled")
+	}
+
 	s := &Server{
-		router:      chi.NewRouter(),
-		readyCtx:    readyCtx,
-		withOTEL:    cfg.GetTelemetryEnabled(),
-		config:      cfg,
-		validator:   validator,
-		userRepo:    userRepo,
-		userService: userService,
-		startedAt:   time.Now(),
+		router:       chi.NewRouter(),
+		readyCtx:     readyCtx,
+		withOTEL:     cfg.GetTelemetryEnabled(),
+		config:       cfg,
+		validator:    validator,
+		userRepo:     userRepo,
+		userService:  userService,
+		cacheService: cacheService,
+		startedAt:    time.Now(),
 	}
 	s.setupRoutes()
 	return s
@@ -351,26 +366,49 @@ func (s *Server) handleVersion(w http.ResponseWriter, r *http.Request) {
 		format = "plain" // default format
 	}
 
+	// Check cache if enabled
+	cacheKey := "version:" + format
+	if s.cacheService != nil {
+		if cached, ok := s.cacheService.Get(cacheKey); ok {
+			log.Trace().Str("cache_key", cacheKey).Msg("Cache hit for version")
+			w.Header().Set("Content-Type", "text/plain")
+			if format == "json" {
+				w.Header().Set("Content-Type", "application/json")
+			}
+			w.Write([]byte(cached.(string)))
+			return
+		}
+	}
+
+	// Build response
+	var response string
 	switch format {
 	case "plain":
 		w.Header().Set("Content-Type", "text/plain")
-		w.Write([]byte(version.Short()))
+		response = version.Short()
 	case "full":
 		w.Header().Set("Content-Type", "text/plain")
-		w.Write([]byte(version.Full()))
+		response = version.Full()
 	case "json":
 		w.Header().Set("Content-Type", "application/json")
-		jsonResponse := fmt.Sprintf(`{
+		response = fmt.Sprintf(`{
 			"version": "%s",
 			"commit": "%s",
 			"built": "%s",
 			"go": "%s"
 		}`, version.Version, version.Commit, version.Date, version.GoVersion)
-		w.Write([]byte(jsonResponse))
 	default:
 		w.Header().Set("Content-Type", "text/plain")
-		w.Write([]byte(version.Short()))
+		response = version.Short()
 	}
+
+	// Cache the response for 60 seconds if cache is enabled
+	if s.cacheService != nil {
+		s.cacheService.Set(cacheKey, response, 60*time.Second)
+		log.Trace().Str("cache_key", cacheKey).Msg("Cached version response")
+	}
+
+	w.Write([]byte(response))
 }
 
 // HealthzResponse represents the Kubernetes-style health check response