Tweak 'fat jar' handling to generally improve performance:

- Allow JarURLConnection to throw a static FileNotFoundException when
  loading classes. This exception is thrown many times when attempting
  to load a class and is silently swallowed so there is no point in
  providing the entry name.
- Expose JarFile.getJarEntryData(AsciiBytes) and store AsciiBytes in
  the JarURLConnection. Previously AsciiBytes were created, discarded
  then created again.
- Use EMPTY_JAR_URL for the JarURLConnection super constructor. The URL
  is never actually used so we can improve performance by using a
  constant.
- Extract JarEntryName for possible caching. The jar entry name
  extracted from the URL is now contained in an inner JarEntryName
  class. This could be cached if necessary (although currently it is
  not because no perceivable performance benefit was observed)

Fixes gh-1119

Cache root jar files in the Handler and also store nested jar files in
the JarEntryData.

See gh-1119

Update JarFile to reuse the previously parsed entries when creating
filtered jars. This saves needing to re-scan the underlying file to
recreate a subset of entries.

See gh-1119

Drop JarEntryFilter arguments from the JarFile constructor and the
getNestedJarFile methods. Filtered JarFiles can still be obtained by
using the getFilteredJarFile() method.

This helps simplify the code a little and will make it easier to add
caching.

See gh-1119

Fix the formatting which was messed up in commit ad1636fd. Also
simplify the RabbitProperties.setVirtualHost logic a little

See gh-1128

The autoconfig strips out slashes where necessary to make a valid
hostname

Fixes gh-1128

(Was an attempt to get Cygwin to recognise the interrupt, but it seems
to be doomed so not worth having the extra code in there really.)

In order to solve issues with multiple or incompatible AccessDecisionManagers, the CRaSH integration will now only delegate to an explicit shellAccessDecisionManager bean refs #1121

It' sthe best we can do, but the container seems to still use our error page
(it just can't have the status code that we asked for).

Fixes gh-748

Fixes gh-1125

Adding an Order to the BeanPostProcessor and catching an exception
are enough to get a simple web app with @EnableGlobalMethodSecurity
and JDBC user details running. It actually doesn't solve an underlying
problem in Spring Security, but I'll deal with that separately.

See gh-1115

Update SolrAutoConfiguration to specifically require HttpSolrServer
and CloudSolrServer classes rather than the generic SolrServer.

This prevents the rather cryptic "@ConditionalOnMissingBean annotations
must specify at least one bean" error that can occur if an earlier
version of solr happens to be on the classpath (as is the case when
hibernate-search is used).

Fixes gh-1098

Refine auto-configuration for Spring Social to:

- Only auto-configure FB/Twitter/LinkedIn if the `app-id` property is
  set.
- Only configure ConnectController and ProviderSignInController if
  there is a ConnectionFactoryLocator.
- Auto-configure Spring Social's SpringSocialDialect for Thymeleaf if
  Thymeleaf is present.
- Added several tests around Spring Social auto-configuration.

Fixes gh-1118

The default behaviour doesn't change with this commit, but now
the user has the option to specify a 'classifier' property
either in springBoot { classifier = 'exec' } (i.e. globally
for all repackage tasks) or in each repackage task, e.g.
bootRepackage { classifier = 'exec' }. In that case the original
archive is not overwritten but copied into <file>-<classifier>.jar
(or .war etc.) and then enhanced.

Fixes gh-1113, fixes gh-141 also I believe.

refs #1121

It needs to run as soon as the DataSource is available really otherwise
anything else that depends on the DataSource (like Security JDBC
initializers) might fail when it tries to use it.

One change from 1.1.1 is that if you have a schema.sql you had better
make sure your data.sql talks to the same tables. In 1.1.1 you could
sometimes get away with letting Hibernate initialize the tables for
your data.sql and *also* have a schema.sql. This was fragile and doomed
to fail eventually if the DataSourceInitializer somehow got
initialized earlier (e.g. through a @DependsOn), so in the spririt
of honesty being the best policy we explicitly disallow it now.

Fixes gh-1115

The class Compile was deprecated and will be removed in Gradle 2.0 (it is
already removed in the release canidate).

Instead we should use JavaCompile. Fixes gh-1066.

See gh-1115

Move ApplicationContextTestUtils from src/main to src/test since it
is probably of limited use to others.

Fixes gh-1034

Prior to this commit it was not safe to start several contexts
using the HornetQAutoConfiguration in the same VM. Each context
was trying to start their own HornetQ embedded broker by default but
only the first was really starting. Worse, the various InVM connection
factories were all silently connecting to the first broker.

This commit introduces a new "serverId" property that is an auto-
incremented integer by default. This identifies the server to connect
to and allows each context to start its own embedded broker in total
isolation of other contexts.

This commits makes it possible for a context to disable its own
embedded broker and connect to an existing one, potentially started
by another context.

Fixes gh-1063

This commit reuses the standard behavior provided by Properties for
parsing keys rather than parsing the keys ourselves. This provides a
consistent features set regardless of where the property is defined.

Fixes gh-1110

Fixes gh-1067

See gh-1072

Fixes gh-320, since this is the officially supported mechanism
for changing the version of Spring (and other dependencies).

Fixes gh-1068