checkscript() wrongly encodes non-ascii text

When `content` takes more bytes than chars, the following command wrongly encodes text:
```python
        content = tools.to_bytes(
            u"{%d+}%s%s" % (len(content), str(CRLF), content))
```
This code works correctly:
```python
        content = tools.to_bytes(content)
        len1 = tools.to_bytes(u"{%d+}" % len(content))
        content = len1 + CRLF + content
```