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/deepseek.go

@@ -3,15 +3,15 @@ package providers
 import (
 	"bufio"
 	"context"
-	"time"
 	"encoding/json"
 	"fmt"
 	"io"
 	"strings"
+	"time"
 
+	"github.com/go-resty/resty/v2"
 	"gophergate/internal/config"
 	"gophergate/internal/models"
-	"github.com/go-resty/resty/v2"
 )
 
 type DeepSeekProvider struct {
@@ -33,11 +33,11 @@ func (p *DeepSeekProvider) Name() string {
 }
 
 type deepSeekUsage struct {
-	PromptTokens          uint32 `json:"prompt_tokens"`
-	CompletionTokens      uint32 `json:"completion_tokens"`
-	TotalTokens           uint32 `json:"total_tokens"`
-	PromptCacheHitTokens  uint32 `json:"prompt_cache_hit_tokens"`
-	PromptCacheMissTokens uint32 `json:"prompt_cache_miss_tokens"`
+	PromptTokens            uint32 `json:"prompt_tokens"`
+	CompletionTokens        uint32 `json:"completion_tokens"`
+	TotalTokens             uint32 `json:"total_tokens"`
+	PromptCacheHitTokens    uint32 `json:"prompt_cache_hit_tokens"`
+	PromptCacheMissTokens   uint32 `json:"prompt_cache_miss_tokens"`
 	CompletionTokensDetails *struct {
 		ReasoningTokens uint32 `json:"reasoning_tokens"`
 	} `json:"completion_tokens_details"`
@@ -75,7 +75,7 @@ func (p *DeepSeekProvider) ChatCompletion(ctx context.Context, req *models.Unifi
 		delete(body, "top_p")
 		delete(body, "presence_penalty")
 		delete(body, "frequency_penalty")
-		
+
 		if msgs, ok := body["messages"].([]interface{}); ok {
 			for _, m := range msgs {
 				if msg, ok := m.(map[string]interface{}); ok {
@@ -142,7 +142,7 @@ func (p *DeepSeekProvider) ChatCompletionStream(ctx context.Context, req *models
 		delete(body, "top_p")
 		delete(body, "presence_penalty")
 		delete(body, "frequency_penalty")
-		
+
 		if msgs, ok := body["messages"].([]interface{}); ok {
 			for _, m := range msgs {
 				if msg, ok := m.(map[string]interface{}); ok {
@@ -175,7 +175,7 @@ func (p *DeepSeekProvider) ChatCompletionStream(ctx context.Context, req *models
 	}
 
 	ch := make(chan *models.ChatCompletionStreamResponse)
-	
+
 	go func() {
 		defer close(ch)
 		// Custom scanner loop to handle DeepSeek specific usage in chunks
@@ -219,3 +219,7 @@ func StreamDeepSeek(ctx io.ReadCloser, ch chan<- *models.ChatCompletionStreamRes
 	}
 	return scanner.Err()
 }
+
+func (p *DeepSeekProvider) ImageGeneration(ctx context.Context, req *models.ImageGenerationRequest) (*models.ImageGenerationResponse, error) {
+	return nil, fmt.Errorf("deepseek does not support image generation")
+}