feat(channel): 缓存扁平化 + 网关映射集成 + 计费模式统一 + 模型限制

- 缓存重构为 O(1) 哈希结构 (pricingByGroupModel, mappingByGroupModel)
- 渠道模型映射接入网关流程 (Forward 前应用, a→b→c 映射链)
- 新增 billing_model_source 配置 (请求模型/最终模型计费)
- usage_logs 新增 channel_id, model_mapping_chain, billing_tier 字段
- 每种计费模式统一支持默认价格 + 区间定价
- 渠道模型限制开关 (restrict_models)
- 分组按平台分类展示 + 彩色图标
- 必填字段红色星号 + 模型映射 UI
- 去除模型通配符支持

backend/internal/service/usage_log.go

@@ -104,6 +104,12 @@ type UsageLog struct {
 	// UpstreamModel is the actual model sent to the upstream provider after mapping.
 	// Nil means no mapping was applied (requested model was used as-is).
 	UpstreamModel *string
+	// ChannelID 渠道 ID
+	ChannelID *int64
+	// ModelMappingChain 模型映射链，如 "a→b→c"
+	ModelMappingChain *string
+	// BillingTier 计费层级标签（per_request/image 模式）
+	BillingTier *string
 	// ServiceTier records the OpenAI service tier used for billing, e.g. "priority" / "flex".
 	ServiceTier *string
 	// ReasoningEffort is the request's reasoning effort level.