Test for LoadString

[Dmitry Timoshkov]

"Vitaly Lipatov" <lav at etersoft.ru> wrote:

> Add tests for LoadString, completed on Windows.
> 
> As I see,
> 1. LoadStringA with buf = NULL segfaults in Windows, we have to produce ERR 
> message instead
> 2. LoadStringA with buf = NULL, buflen = 0 returns -1
> 3. LoadStringW with some buf, buflen = 0 returns length of the resource string 
> and puts some strange chars (F%O f.i.) in the buffer.
> 4. LoadStringW with buf = NULL, buflen!=0 returns 0
> 5. LoadStringA/W segfaults on negative buflen

Unfortunately the tests don't show whether last error value should be set
in the case of a failure or not, please add that as well.

-- 
Dmitry.