DNF90 服务端与启动器完整源码（含 GM 工具功能增强）

服务端（Go）：
- GM 塞背包/带属性装备发放：在线角色推送 op14 增量物品行，背包实时刷新，无需回角色选择界面
- GM 改金币/改等级：去掉在线拒绝，落库后推送 op14 钱包/op37 等级状态，在线实时刷新
- 等级端点统一切换到新版 GMSetCharacterLevel（自带在线刷新）
- 无限/锁死疲劳模式 + 燃烧疲劳：每房间消耗 2 点、打怪与通关经验双倍
- 燃烧疲劳开关时给所有在线角色推送系统提示（op12 mode 0）
- 深渊史诗掉率：GM 全局倍率 + 城镇增益药剂窗口倍率，真实接入掉落结算

启动器（Python/JS）：
- 标题栏自绘、窗口 1204x717、禁止拖拽改大小、仅最小化/关闭
- GM 工具：疲劳与燃烧、深渊史诗掉率、拍卖行上架（单一定价对齐）
- 拍卖行上架前端对齐服务端单一价格语义，修复 400 报错

server/internal/services/dnfbridge/account_adventure_constants.go

@@ -0,0 +1,57 @@
+package dnfbridge
+
+import (
+	"longheng.io/server/internal/modules/dnf/adventuregroup"
+	"longheng.io/server/internal/modules/dnf/dnfenum"
+)
+
+const (
+	currentAdventureInfoPushMsgID               = uint16(dnfenum.CmdPacketCustomAbilityEquipOption)
+	currentAdventureActorRefreshMsgID           = uint16(dnfenum.CmdPacketAgitWarMissionReward)
+	currentAdventureInfoRequestWireLength       = 16
+	currentAdventureInfoSceneRequestLength      = 2
+	currentAdventureInfoRawLength               = 7420
+	currentAdventureActorRefreshRawLength       = 16
+	currentAdventureInfoManageLevelOffset       = 0
+	currentAdventureInfoCurrentPointOffset      = 8
+	currentAdventureInfoLoginDaysOffset         = 20
+	currentAdventureInfoHighestHonorJobOffset   = 44
+	currentAdventureInfoHighestHonorGrowOffset  = 45
+	currentAdventureInfoHighestHonorLevelOffset = 48
+	currentAdventureInfoRosterOffset            = 60
+	currentAdventureInfoRosterCount             = 100
+	currentAdventureInfoRosterEntrySize         = 52
+	currentAdventureInfoRosterHonorOffset       = 4
+	currentAdventureInfoRosterNameOffset        = 12
+	currentAdventureInfoRosterNameSize          = 30
+	currentAdventureInfoRosterRegionOffset      = 42
+	currentAdventureInfoRosterCardLevelOffset   = 48
+	currentAdventureInfoCharacterCountOffset    = 56
+	currentAdventureInfoShopPointOffset         = 5260
+	currentAdventureInfoShopPointCount          = 5
+	currentAdventureInfoShopPointEntrySize      = 8
+	currentAdventureInfoPurchaseOffset          = 5300
+	currentAdventureInfoPurchaseCount           = 256
+	currentAdventureInfoPurchaseEntrySize       = 8
+	currentAdventureInfoTripleOffset            = 7348
+	currentAdventureInfoTripleCount             = 20
+	currentAdventureInfoTripleSize              = 3
+	// Each native row is type:u8 then count:u16. The order matters: the client
+	// reads the classification byte first (see the native regression test).
+	currentAdventureInfoTripleTypeOffset    = 0
+	currentAdventureInfoTripleCountOffset   = 1
+	currentAdventureInfoGrowthCapsuleOffset = 7408
+	// currentAdventureInfoCapsuleClaimCountOffset 是成长胶囊页签"可获得的经验值"
+	// 那一栏的值，也就是**现在可以领取几次**（每次 = 10% 档 = 1,000,000 点）。
+	// 客户端解析 sub_C7DAF0 时把这个 dword 存进 manager+0x310，而
+	// 0x1DE1F30 = `mov eax,[ecx+0x310]; ret` —— 反汇编直接确认。
+	// 它同时被 0x13F0FA0 当成循环上限逐个建列表项，所以**只能发次数**，
+	// 不能发经验值总量（发 1,100 万会让客户端建上千万个字符串而卡死）。
+	currentAdventureInfoCapsuleClaimCountOffset = 7416
+	currentSelectorAdventureInfoSlotMetadata    = adventuregroup.SelectorSlotMetadataKey
+)
+
+// currentAdventureInfoContentCountOffsets are the six u32 content counters.
+// The client reads them at exactly these offsets; the order is the native
+// index order (ContentCounts[0] -> 16, [1] -> 32, ...), not sorted ascending.
+var currentAdventureInfoContentCountOffsets = [adventuregroup.ContentTypeCount]int{16, 32, 24, 28, 36, 40}