Class packaging information is now automatically added to stack traces formatted by PatternLayout

Not using Geronimo JMS instead of Sun JMS to avoid manual installation

this change reduces the probability of that happening

This code attempts to guarantee that the extracted class packaging information 
is extracting using the *exact* same classloader was used to load a given class
in the stack frame. 

Associated test cases. 

 	 
 	 Working on the reliability of the extracted package information which is actually quite a difficult problem.

Reworking Throwable to string conversion. Instead of simply converting StackTraceElement 
(STE) array into just strings, we convert them to a little more sophisticated objects,
namely ThrowableDataPoints which support PackageInformation.

All tests pass.

Reworking Throwable to string conversion. Instead of simply converting StackTraceElement 
(STE) array into just strings, we convert them to a little more sophisticated objects,
namely ThrowableDataPoints which support PackageInformation.

This is ongoing work, unit test may not pass. 

  
  Unfortunately, this implementations fails the Scenario unit tests. Yes, it looks 
  like a JDK bug (very suprisingly).
  
  Commiting this version for historical (archiving) purposes.

Extracting package information is a time consuming process. 
Improve performance by keeping previously found results in a cache.
Added an LRUCache with accompanying test cases for this purpose. 

The c.q.l.core.util.Loader class is used to load classes or locate resources. 
Previously, the TCCL (Thread Context Class Loader) was used to locate resources,
in particular configuration files. However, the TCCL is not set in all environments. 

Instead of using the TCL use now use the class loader that loaded the Loader class
itself to load resources. Now, under many circumstances TCCL==self classloader
so most users will not see a difference. The situation where TCCL != self cl
can occur under relatively complicated scenarios. Thus, this commit simplifies the code
without hopefully affecting existing deployments. 

added a small ant script to automatically copy slf4j-api-${VERSION}.jar to lib/

- Fixes both LBCORE-27 and LBCLASSIC-67. 

- added test cases (using GreenMail) that verify that the output generated
  by SMTPAppender and HTMLLayout conform to xhtml1-strict.dtd

- HTMLLayout* now use StringBuilder instead of StringBuffer. The former is significantly faster.

Other missing files

Missing file

Not not all PatternLayout instances need to have a
ThrowableInformationConverter added at the end of the converter chain
by the postCompileProcessing() method.

With this commit, post compile processing code has been moved to an
interface. By default, PatternLayouts that ship with logback have a
default post compile processor. However, it is now very easy to set
processor to null.

We now test SMTPAppender using GreenMail. Problem described in
LBCLASSIC-67 is reproduced in tests but not yet fixed.

Documentation improvements many of which have been proposed by Anton Tagunov.

Add references to SLF4J in the manual, so that new users have an easier time to find it.

  Chapter3 has been split into two parts, the fist part on logback configuration and a second part on Joran as a standalone framework

- ExecutionContext class has been renamed as InterpretationContext a few months ago. Syncing the docs.
- ongoing work on the documentation

  The previous syntax is still honored but it will no longer be documented.

- ongoing work on the docs

really skip over children when an error occurs. In the previous version of the code, 
we would skip over invoking the action but would still try to find actions for the event. 
In this revision, we skip over the event as a whole, in particular we no longer look up
actions for it. In practical terms, this just avoids adding error statuses for not found 
implicit actions.