feat: Ollama adapter + chat rate limiter (30 req/hour)

Ollama adapter (internal/chat/ollama.go):
- Implements model.LLM interface for ADK Go
- Talks to Ollama's OpenAI-compatible API (/v1/chat/completions)
- Full tool/function calling support (tested with Mistral Small 3.2)
- Converts ADK types to OpenAI format (messages, tools, tool_calls)
- Configurable via OLLAMA_HOST and OLLAMA_MODEL env vars

Multi-provider handler:
- MODEL_PROVIDER env: "gemini" (default) or "ollama"
- Gemini: requires GOOGLE_API_KEY (pay-as-you-go recommended)
- Ollama: connects to local or Tailscale-remote instance

Rate limiter:
- 30 requests/hour per IP on /api/chat endpoint
- Uses existing middleware.NewRateLimiter pattern

Tested: Ollama + Mistral Small 3.2 on M4 Pro 64GB — correct answers

internal/chat/handler.go

@@ -13,6 +13,7 @@ import (
 	"github.com/juanatsap/cv-site/internal/cache"
 
 	"google.golang.org/adk/agent"
+	"google.golang.org/adk/model"
 	"google.golang.org/adk/model/gemini"
 	"google.golang.org/adk/runner"
 	"google.golang.org/adk/session"
@@ -26,26 +27,28 @@ type Handler struct {
 	enabled        bool
 }
 
-// NewHandler creates a chat handler. Returns a disabled handler if GOOGLE_API_KEY is not set.
+// NewHandler creates a chat handler. Returns a disabled handler if no model provider is configured.
 func NewHandler(dataCache *cache.DataCache) *Handler {
-	apiKey := os.Getenv("GOOGLE_API_KEY")
-	if apiKey == "" {
-		log.Println("⚠️  GOOGLE_API_KEY not set — chat feature disabled")
-		return &Handler{enabled: false}
+	provider := os.Getenv("MODEL_PROVIDER")
+	if provider == "" {
+		provider = "gemini"
 	}
 
-	ctx := context.Background()
+	var llm model.LLM
+	var providerLabel string
 
-	modelName := os.Getenv("MODEL_NAME")
-	if modelName == "" {
-		modelName = "gemini-2.5-flash"
+	switch provider {
+	case "ollama":
+		llm, providerLabel = initOllamaProvider()
+	default:
+		var err error
+		llm, providerLabel, err = initGeminiProvider()
+		if err != nil {
+			return &Handler{enabled: false}
+		}
 	}
 
-	llm, err := gemini.NewModel(ctx, modelName, &genai.ClientConfig{
-		APIKey: apiKey,
-	})
-	if err != nil {
-		log.Printf("⚠️  Failed to initialize Gemini model: %v — chat disabled", err)
+	if llm == nil {
 		return &Handler{enabled: false}
 	}
 
@@ -68,7 +71,7 @@ func NewHandler(dataCache *cache.DataCache) *Handler {
 		return &Handler{enabled: false}
 	}
 
-	log.Printf("💬 Chat agent enabled (model: %s)", modelName)
+	log.Printf("💬 Chat agent enabled (%s)", providerLabel)
 
 	return &Handler{
 		runner:         r,
@@ -77,6 +80,48 @@ func NewHandler(dataCache *cache.DataCache) *Handler {
 	}
 }
 
+// initGeminiProvider initializes the Gemini LLM provider.
+func initGeminiProvider() (model.LLM, string, error) {
+	apiKey := os.Getenv("GOOGLE_API_KEY")
+	if apiKey == "" {
+		log.Println("⚠️  GOOGLE_API_KEY not set — chat feature disabled")
+		return nil, "", fmt.Errorf("no API key")
+	}
+
+	ctx := context.Background()
+
+	modelName := os.Getenv("MODEL_NAME")
+	if modelName == "" {
+		modelName = "gemini-2.5-flash"
+	}
+
+	llm, err := gemini.NewModel(ctx, modelName, &genai.ClientConfig{
+		APIKey: apiKey,
+	})
+	if err != nil {
+		log.Printf("⚠️  Failed to initialize Gemini model: %v — chat disabled", err)
+		return nil, "", err
+	}
+
+	return llm, fmt.Sprintf("gemini: %s", modelName), nil
+}
+
+// initOllamaProvider initializes the Ollama LLM provider.
+func initOllamaProvider() (model.LLM, string) {
+	host := os.Getenv("OLLAMA_HOST")
+	if host == "" {
+		host = "http://localhost:11434"
+	}
+
+	modelName := os.Getenv("OLLAMA_MODEL")
+	if modelName == "" {
+		modelName = "mistral-small3.2"
+	}
+
+	llm := NewOllamaModel(host, modelName)
+	return llm, fmt.Sprintf("ollama: %s @ %s", modelName, host)
+}
+
 // Enabled returns whether the chat feature is available.
 func (h *Handler) Enabled() bool {
 	return h.enabled