Initialization test

[Maxim Kizub]

>I'm sorry if you will get multiple copies of that mail.
>
>Attached is test for class initialization. I know that it might be hard
>to conform to it, but specs say so. I know for sure that tya fails this

 Whould you be so kind to expand word "specs"?

At least, I have these notes to the test:

1) the results may depend not only on JVM/JIT, but on code generated
by compiler. See test for Class4.
2) to pass this test JIT should compile the code in inefficient way. See
next note.
3) dynamic class loading rules are not clear enough for programmers
to belive them in any situation. This way of programing is *bad style*.

>test, I haven't checked kaffe and japhar, because I do not have them at
>hand.
>
>Feel free to include it into your regression testing if you do not have
>something similar yet.