Fix: Decode process list names/paths as GBK on Windows

Process management decoded the client's process name/path bytes with
GetClientEncoding, which returns UTF-8 for CLIENT_CAP_UTF8 clients. But
the Windows client builds the process list with the ANSI A-interface
(Process32FirstA / QueryFullProcessImageNameA), which emits the client's
system code page (GBK on Chinese Windows) regardless of the UTF-8
capability -- unlike window titles, which were migrated to the
W-interface plus UTF-8. Decoding GBK as UTF-8 garbled Chinese process
names and paths.

Add GetProcessListEncoding(): 936 for Windows clients, UTF-8 for
LNX/MAC, matching McpServer::BuildListProcesses, and use it in
ShowProcessList (which still writes via LVM_SETITEMTEXTW).

Also raise the IP history cap to 1000 and rename the experimental
"隐私屏幕" menu item to "虚拟屏幕".

Co-Authored-By: deepseek-v4-pro

server/2015Remote/SystemDlg.cpp

@@ -232,6 +232,26 @@ void CSystemDlg::ShowWindowsList(void)
     ApplySort();  // 应用记忆的排序偏好
 }
 
+// 按客户端编码 (936 或 UTF-8) 解码字节串，并用 LVM_SETITEMTEXTW 写入列表单元格，
+// 绕开 ANSI→CP_ACP 回转（非中文 ACP 服务端上中文也能正常显示，与窗口管理同法）。
+static void SetClientTextItem(CListCtrl& list, int iItem, int iSubItem, const char* sz, UINT cp)
+{
+    std::wstring w;
+    if (sz && sz[0]) {
+        int wlen = MultiByteToWideChar(cp, 0, sz, -1, NULL, 0);
+        if (wlen > 1) {
+            w.resize(wlen - 1);
+            MultiByteToWideChar(cp, 0, sz, -1, &w[0], wlen);
+        }
+    }
+    LVITEMW lv = {};
+    lv.mask = LVIF_TEXT;
+    lv.iItem = iItem;
+    lv.iSubItem = iSubItem;
+    lv.pszText = w.empty() ? const_cast<LPWSTR>(L"") : &w[0];
+    ::SendMessageW(list.GetSafeHwnd(), LVM_SETITEMTEXTW, (WPARAM)iItem, (LPARAM)&lv);
+}
+
 void CSystemDlg::ShowProcessList(void)
 {
     // GetMyBuffer(1) 深拷贝数据，跳过 TOKEN 字节
@@ -248,6 +268,10 @@ void CSystemDlg::ShowProcessList(void)
     const char	*szProcessFullPath;
     DWORD	dwOffset = 0;
     CString str;
+    // 进程名/路径编码：Windows 客户端 A 接口生成，恒为 GBK/936；仅 LNX/MAC 为 UTF-8。
+    // 不能用 GetClientEncoding（它对新客户端返回 CP_UTF8，会把 GBK 进程名解成乱码）。
+    UINT cp = GetProcessListEncoding(m_ContextObject);
+
     DeleteAllItems();
     //遍历发送来的每一个字符别忘了他的数据结构啊 Id+进程名+0+完整名+0
     int i;
@@ -258,10 +282,11 @@ void CSystemDlg::ShowProcessList(void)
         szProcessFullPath = szExeFile + lstrlen(szExeFile) + 1;  //完整名就是进程名之后的啦
         //他的数据结构的构建很巧妙
 
-        m_ControlList.InsertItem(i, arr[0].c_str());       //将得到的数据加入到列表当中
+        m_ControlList.InsertItem(i, "");       // 占位，首列随后用 Unicode 写入
         str.FormatL("%5u", *PID);
-        m_ControlList.SetItemText(i, 1, str);
-        m_ControlList.SetItemText(i, 2, szProcessFullPath);
+        m_ControlList.SetItemText(i, 1, str);              // PID (ASCII)
+        SetClientTextItem(m_ControlList, i, 0, arr[0].c_str(), cp);      // 进程名（按客户端编码）
+        SetClientTextItem(m_ControlList, i, 2, szProcessFullPath, cp);   // 程序路径（按客户端编码）
         m_ControlList.SetItemText(i, 3, arr[1].empty() ? "N/A" : arr[1].c_str());
         // ItemData 为进程ID；Data[3] 需与第 4 列"架构"一致，否则按架构排序拿到空串而无意义
         auto data = new ItemData{ *PID, {arr[0].c_str(), str, szProcessFullPath, arr[1].empty() ? "N/A" : arr[1].c_str()}, arr[1].c_str() };