feat(rpm): RPM 限流模块优化

P0:
- rpm_override 嵌入 Auth Cache Snapshot，消除每请求 DB 查询 (snapshot v6→v7)
- 429 RPM 响应返回 Retry-After 头（当前分钟剩余秒数）

P1:
- ClearAll 按钮直连 DELETE API，带 loading 防重复
- 新增 GET /admin/users/:id/rpm-status 管理员 RPM 用量查询端点

优化:
- checkRPM 从级联互斥改为并行取最严，user.rpm_limit 作为全局硬上限始终生效
- Override/Group 变更后自动失效 auth cache
- fail-open 语义不变，Redis 故障不阻塞业务

backend/ent/group/group.go

@@ -76,6 +76,8 @@ const (
 	FieldDefaultMappedModel = "default_mapped_model"
 	// FieldMessagesDispatchModelConfig holds the string denoting the messages_dispatch_model_config field in the database.
 	FieldMessagesDispatchModelConfig = "messages_dispatch_model_config"
+	// FieldRpmLimit holds the string denoting the rpm_limit field in the database.
+	FieldRpmLimit = "rpm_limit"
 	// EdgeAPIKeys holds the string denoting the api_keys edge name in mutations.
 	EdgeAPIKeys = "api_keys"
 	// EdgeRedeemCodes holds the string denoting the redeem_codes edge name in mutations.
@@ -181,6 +183,7 @@ var Columns = []string{
 	FieldRequirePrivacySet,
 	FieldDefaultMappedModel,
 	FieldMessagesDispatchModelConfig,
+	FieldRpmLimit,
 }
 
 var (
@@ -258,6 +261,8 @@ var (
 	DefaultMappedModelValidator func(string) error
 	// DefaultMessagesDispatchModelConfig holds the default value on creation for the "messages_dispatch_model_config" field.
 	DefaultMessagesDispatchModelConfig domain.OpenAIMessagesDispatchModelConfig
+	// DefaultRpmLimit holds the default value on creation for the "rpm_limit" field.
+	DefaultRpmLimit int
 )
 
 // OrderOption defines the ordering options for the Group queries.
@@ -403,6 +408,11 @@ func ByDefaultMappedModel(opts ...sql.OrderTermOption) OrderOption {
 	return sql.OrderByField(FieldDefaultMappedModel, opts...).ToFunc()
 }
 
+// ByRpmLimit orders the results by the rpm_limit field.
+func ByRpmLimit(opts ...sql.OrderTermOption) OrderOption {
+	return sql.OrderByField(FieldRpmLimit, opts...).ToFunc()
+}
+
 // ByAPIKeysCount orders the results by api_keys count.
 func ByAPIKeysCount(opts ...sql.OrderTermOption) OrderOption {
 	return func(s *sql.Selector) {

backend/ent/group/where.go

@@ -190,6 +190,11 @@ func DefaultMappedModel(v string) predicate.Group {
 	return predicate.Group(sql.FieldEQ(FieldDefaultMappedModel, v))
 }
 
+// RpmLimit applies equality check predicate on the "rpm_limit" field. It's identical to RpmLimitEQ.
+func RpmLimit(v int) predicate.Group {
+	return predicate.Group(sql.FieldEQ(FieldRpmLimit, v))
+}
+
 // CreatedAtEQ applies the EQ predicate on the "created_at" field.
 func CreatedAtEQ(v time.Time) predicate.Group {
 	return predicate.Group(sql.FieldEQ(FieldCreatedAt, v))
@@ -1320,6 +1325,46 @@ func DefaultMappedModelContainsFold(v string) predicate.Group {
 	return predicate.Group(sql.FieldContainsFold(FieldDefaultMappedModel, v))
 }
 
+// RpmLimitEQ applies the EQ predicate on the "rpm_limit" field.
+func RpmLimitEQ(v int) predicate.Group {
+	return predicate.Group(sql.FieldEQ(FieldRpmLimit, v))
+}
+
+// RpmLimitNEQ applies the NEQ predicate on the "rpm_limit" field.
+func RpmLimitNEQ(v int) predicate.Group {
+	return predicate.Group(sql.FieldNEQ(FieldRpmLimit, v))
+}
+
+// RpmLimitIn applies the In predicate on the "rpm_limit" field.
+func RpmLimitIn(vs ...int) predicate.Group {
+	return predicate.Group(sql.FieldIn(FieldRpmLimit, vs...))
+}
+
+// RpmLimitNotIn applies the NotIn predicate on the "rpm_limit" field.
+func RpmLimitNotIn(vs ...int) predicate.Group {
+	return predicate.Group(sql.FieldNotIn(FieldRpmLimit, vs...))
+}
+
+// RpmLimitGT applies the GT predicate on the "rpm_limit" field.
+func RpmLimitGT(v int) predicate.Group {
+	return predicate.Group(sql.FieldGT(FieldRpmLimit, v))
+}
+
+// RpmLimitGTE applies the GTE predicate on the "rpm_limit" field.
+func RpmLimitGTE(v int) predicate.Group {
+	return predicate.Group(sql.FieldGTE(FieldRpmLimit, v))
+}
+
+// RpmLimitLT applies the LT predicate on the "rpm_limit" field.
+func RpmLimitLT(v int) predicate.Group {
+	return predicate.Group(sql.FieldLT(FieldRpmLimit, v))
+}
+
+// RpmLimitLTE applies the LTE predicate on the "rpm_limit" field.
+func RpmLimitLTE(v int) predicate.Group {
+	return predicate.Group(sql.FieldLTE(FieldRpmLimit, v))
+}
+
 // HasAPIKeys applies the HasEdge predicate on the "api_keys" edge.
 func HasAPIKeys() predicate.Group {
 	return predicate.Group(func(s *sql.Selector) {