feat: add image generation for OpenAI DALL-E and Gemini Imagen

New `/v1/images/generations` endpoint proxies DALL-E 2/3 (OpenAI)
and Imagen 3 (Gemini). Same auth/logging as chat completions.

- Add ImageGenerationRequest/Response models
- Extend Provider interface with ImageGeneration()
- OpenAI: forward to /v1/images/generations
- Gemini: call /v1beta/models/{model}:predict, map OpenAI params
- Circuit breaker wraps image gen like chat completions
- Model routing: dall-e* -> openai, imagen*/gemini* -> gemini
- Unsupported providers (deepseek/moonshot/grok/ollama) return error
- Fix pre-existing CachedContentTokenCount bug in StreamGemini

internal/providers/ollama.go

@@ -249,3 +249,7 @@ func StreamOllama(ctx io.ReadCloser, ch chan<- *models.ChatCompletionStreamRespo
 	}
 	return scanner.Err()
 }
+
+func (p *OllamaProvider) ImageGeneration(ctx context.Context, req *models.ImageGenerationRequest) (*models.ImageGenerationResponse, error) {
+	return nil, fmt.Errorf("ollama does not support image generation")
+}