fix: sanitize archive entry paths before extraction to prevent path traversal

Move path sanitization before directory handling in extractFile so
directory entries with "../" prefixes are rejected the same way file
entries already are. Add the same sanitization to extractSevenZipMultiPart
and both zip extraction loops in the Windows updater, which had no path
validation at all.

cmd/updater/updater_windows.go

@@ -37,15 +37,17 @@ func installByInstaller(killSignalChan chan<- any, packagePath, destDir string)
 	}
 	defer reader.Close()
 
-	// Find the installer file
 	var installerPath string
 	for _, file := range reader.File {
+		cleanName := filepath.Clean(file.Name)
+		if strings.HasPrefix(cleanName, "..") || filepath.IsAbs(cleanName) {
+			continue
+		}
 		if file.FileInfo().IsDir() {
 			continue
 		}
 
-		// Extract file
-		path := filepath.Join(tempDir, file.Name)
+		path := filepath.Join(tempDir, cleanName)
 
 		if err := os.MkdirAll(filepath.Dir(path), os.ModePerm); err != nil {
 			return err
@@ -101,7 +103,11 @@ func installByPortable(killSignalChan chan<- any, packagePath, destDir string) e
 	defer reader.Close()
 
 	for _, file := range reader.File {
-		path := filepath.Join(destDir, file.Name)
+		cleanName := filepath.Clean(file.Name)
+		if strings.HasPrefix(cleanName, "..") || filepath.IsAbs(cleanName) {
+			continue
+		}
+		path := filepath.Join(destDir, cleanName)
 
 		if file.FileInfo().IsDir() {
 			os.MkdirAll(path, file.Mode())