LOGBACK-833 - allow previously defined properties (in local scope) to be seen by SiftingJoranConfigurator
 * implemented part two of Ceki's suggestion
 * add getCopyOfPropertyMap() to the PropertyContainer and thus InterpretationContext
 * use local properties that are ineffect at time of creation of SiftingAppender to seed the properties map used when creting the child appenders and properties
 * removed some generics warnings
 * in AppenderFactoryBase use List.subList() rather than "copy whole array then delete and re-shuffle elements"

LOGBACK-835 - make timestamp properties default to local scope. Can be overriden by setting scope attribute

make sure that all the 'view as groovy' links work, Gaffer does not support <receiver> at this stage, nor does the translator

This should resolve LOGBACK-749.

throws an exception, attempt to provide the IP address instead.  This
is useful on those systems without a hostname explicitly defined.

made FileAppenderResilienceTest more tolerant, added code wait for compression job in TimeBasedRollingPolicy

In the receiveMessage and receiveWithContext tests, only ten
milliseconds were allowed for the appender to dispatch the message over
the socket.  Practical experience suggests that on the build system, the
actual latency is on the order of hundreds of milliseconds.  I changed
the allowed wait time to 1000 milliseconds.  This should have little
impact on test duration, as the test will finish as soon as the latch
we're awaiting is updated.

The number of iterations for the loop in lateServerLaunch was derived
from the delay allowed for the appender latch.  The new value for the
delay resulted in an iteration count of one.  I changed this test to
simply use a fixed iteration count.

testConnectionFails was expecting the connector thread to exit after
being interrupted, with latency no greater than the configured delay
between connection attempts.  However, the call to
SocketFactory.createSocket may not return immediately on interrupt.  

Now, we allow 4 * DELAY for the thread to exit. This commit should allow
more than enough tolerance, but (as it generally has no effect on the
run duration of the test) we could increase the allowed delay even
further if it proves too sensitive to timing variations on the build
system.