refactor(idgen): centralize negative ID handling via panic

Restore NextUint64ID() to uint64-only API so callers need no error
checks. Retry logic stays in idgen; after 3 negative values it panics
instead of returning 0, preventing silent NULL primary keys.

internal/db/idgen/snowflake.go

@@ -6,7 +6,6 @@
 package idgen
 
 import (
-	"errors"
 	"fmt"
 	"log"
 
@@ -19,9 +18,6 @@ const epoch int64 = 1764547200000
 
 const maxNegativeIDRetries = 3
 
-// ErrNegativeSnowflakeID 表示 Snowflake 在重试后仍生成负值 ID。
-var ErrNegativeSnowflakeID = errors.New("snowflake generated negative ID")
-
 var node *snowflake.Node
 
 func init() {
@@ -37,14 +33,14 @@ func init() {
 }
 
 // NextUint64ID 生成下一个分布式唯一 ID。
-// 理论上不应出现负值；若出现则最多重试 maxNegativeIDRetries 次，仍失败则返回错误。
-func NextUint64ID() (uint64, error) {
+// 理论上不应出现负值；若出现则最多重试 maxNegativeIDRetries 次，仍失败则 panic。
+func NextUint64ID() uint64 {
 	for attempt := 1; attempt <= maxNegativeIDRetries; attempt++ {
 		id := node.Generate().Int64()
 		if id >= 0 {
-			return uint64(id), nil
+			return uint64(id)
 		}
 		log.Printf("[Snowflake] generated negative ID: %d (attempt %d/%d)", id, attempt, maxNegativeIDRetries)
 	}
-	return 0, fmt.Errorf("%w: failed after %d attempts", ErrNegativeSnowflakeID, maxNegativeIDRetries)
+	panic(fmt.Sprintf("[Snowflake] generated negative ID after %d attempts", maxNegativeIDRetries))
 }

internal/db/idgen/snowflake_test.go

@@ -7,11 +7,9 @@ import (
 	"testing"
 
 	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
 )
 
 func TestNextUint64ID(t *testing.T) {
-	id, err := NextUint64ID()
-	require.NoError(t, err)
+	id := NextUint64ID()
 	assert.NotZero(t, id)
 }