Thus, the UPLOAD management is much more stable and relies on a well documented
and maintained library (but at the cost of the additionnal size of the final
JAR). The most important point is the capability to efficiently limit the size
of the incoming file and on the disk/memory management when uploading the
file(s) and in case of error during this process (i.e. files must be deleted).

Then, UWS or TAP implementors are free to use whatever logging system they want
(e.g. log4j, logback, ...).

Before, the scripts relied unfortunately on the compilation performed by
Eclipse. Now, all the JUnit test cases are compiled all the time before
performing the JUnit validation (i.e. ANT task named `junitValidation`).

This is very useful when getting a stack trace of an exception: without
this option, no method and line information can be provided...so, without
this option, the stack trace becomes pretty useless.
(NOTE: this option increases a little the size of the compiled classes)

in the log output. For that the function UWSToolBox.getParamsMap(...)
had to be modified.

[TAP] Upgrade STIL version + Specify charset encoding for Javadoc generation in all Ant build scripts.

[ALL] Update ANT build scripts to test all existings JUnit tests before compiling. In case of error, the compilation does not complete.

TAP: Execute all JUnit tests before building the TAP library. At any JUnit failure, the build is canceled
+ Update the version number for the generated JARs.