x.json2: fix `json_string` for utf8 codepoints, add regression test

vlib/x/json2/encoder.v

@@ -156,7 +156,10 @@ fn json_string(s string) string {
 		} else {
 			slice := s[i..i + char_len]
 			hex_code := slice.utf32_code().hex()
-			if hex_code.len == 4 {
+			if hex_code.len < 4 {
+				// an utf8 codepoint
+				sb.write_string(slice)
+			} else if hex_code.len == 4 {
 				sb.write_string('\\u$hex_code')
 			} else {
 				// TODO: still figuring out what

vlib/x/json2/encoder_test.v

@@ -19,3 +19,11 @@ fn test_json_string_non_ascii() {
 	text := json2.Any('ひらがな')
 	assert text.json_str() == r'\u3072\u3089\u304c\u306a'
 }
+
+fn test_utf8_strings_are_not_modified() ? {
+	original := '{"s":"Schilddrüsenerkrankungen"}'
+	// dump(original)
+	deresult := json2.raw_decode(original) ?
+	// dump(deresult)
+	assert deresult.str() == original
+}