[8u] RFR 8233197: Invert JvmtiExport::post_vm_initialized() and Jfr:on_vm_start() start-up order for correct option parsing

[Jaroslav Bachorík]

Please review the following backport

JIRA.     : https://bugs.openjdk.java.net/browse/JDK-8233197
Webrev. : http://cr.openjdk.java.net/~jbachorik/8233197/

The backport patch applied rather cleanly, not considering several offset
adjustments.
The only part that required an additional change was modifying the place
where java lang classes initialization happens in thread.cpp - those
classes need to be initialized before 'Jfr::on_create_vm_2()' is called. In
order to achieve this I just moved around the whole codeblock related to
java lang classes initialization.

All tests from jdk_jfr are passing after this patch has been applied.

Thanks!

-JB-