Changes

Partial cleanup in war-util and classloaders

- generics, stringbuffers, formatting, deprecated methods
- bit of optimization
- debug removed (unused)
- delegate is true by default, synced
- logic around isDelegateFirstResource simplified (probably will be revisited
  for Faces TCK)

Implement Jakarta Security glue code.

Signed-off-by: Arjan Tijms <arjan.tijms@gmail.com>

Faster comparison in ResourceLocator

- if strings differ in length, the result is obtained faster

Issue #23873 Fixed some log messages and added warnings to detect missing bundles

Issue #23873 Fixed ORB logs

- created IIOPImplLogFacade
- some missing keys were found in logs, some were found in sources
- CULoggerInfo contained commented out code - some keys were used on other places

Issue #23873 Fix for quicklook tests - LogDomains vs. bundles

- for some unknown reason tests fail when we use standard of JDK8+ for
  setting a resource bundle to a logger. I couldn't find the exact cause
  of the failure (invalid tx state) now, but I will get back to it later
  after jta/jps refactoring and also with new logging it should be easier.
  That is why I added the FIXME here - and also to prevent me or someone
  else breaking it again.

GJULE integration with GlassFish

- own folder and CL for bootstrap dependencies - no need for JVM options
- sorted logging.properties
- HK2 services ARE NOT the logging system, but USE the logging system
- AdminMain sets the log manager, but doesn't set the configuration yet - this
  step is on the server (it has to load logging.properties first). All log
  records are buffered automatically -> no need for earlyLogMessages any more,
  they also contained around 20 selected messages, but all other logs were lost.

- url classloaders have toString
- DirectoryClassLoader doesn't throw IOE but ISE -> if anything would go wrong,
  we have ISE with stacktrace instead of NPE following swallowed IOE.

- Fixed some log messages
- Fixed SecurityManager complaints in QuickLook tests

Formatting

Fixed printing OSGI issues in EmbeddedOSGiGlassFishRuntime

- originally I even removed the FelixPrettyPrinter usage as I did not see
  any difference, then I noticed that MultiExceptions are sometimes incorrectly
  layered by HK2 and it was possible just to enhance the algorithm for
  finding the BundleException.
- moved and extended the test from main to JUnit.

Brief Product Name removed

- it was redundant when we already have Abbreviated Product Name
- the only difference was it was lowercase, but that did not look well when
  used in logs

Removed unused LogDomains constants and NAMING

- NAMING duplicated JNDI which was used more often

Fixed sorting inside logging.properties

Killing zombies - removed unused classes and constants and misleading comments

Removed unused and obsoleted JavaVersion class

- since JDK9 we have Runtime.version()

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>