Fixed typo

In case you have clustered admin servers, you can now specify multiple urls for
the client to register at. So that in the case the first server is down it will
register at the next one specified in the list. This allows you to do cluster-
ing without the need for a load-balanced host.

closes #137

Remove the git-commit-id-plugin from the sample and don't fail in docs when the
sources are not located in an git-repository (e.g. downloaded from zip)

I did an accidental release of a 1.3.2, so we need another increment

There is a problem with the relaxed binding for system environment variables in
spring boot and ignoreUnknownFields for ConfigurationProperties. So we need to
remove it for now.

fixes #135

Cause it was often requested, now the context-path for the admin server is
configurable. This allows to move the admin-ui and endpoints to other http-
paths than "/".
NOTE: I'd still advise not to add the admin to the application you want to
monitor.

For less commits to the gh-pages remove the last-update-label from the footer.
So that a commit is only made if the content changes.

This should avoid confusion as in #131

Take the servlet-path into account when computing the management-/healht-url.
This is not done for the service-url, since the main use-case for setting the
servlet-path is to get the dispatcher-servlet out of the way for other servlets.

fixes #133

Spring Cloud Brixton hasn't been released yet, so to get the Spring Boot Admin
release out. We downgrade to Spring Cloud Angel.SR3.

In cas a pull request is tested, the secret environment variables are not
present, because they could be exposed by a malicious pull request. So we can
only decrypt the gpg-keys only when the keyphrase is availible.

Rather than relying on a constant mbean name for logbacks JMXConfigurator mbean, search for
all available beans of the type. If a single one is found take that. If multiple are found
first choose the one with Name=application name, if not present take the one with Name=default.
In case none matches display an error.

closes #124

closes #121

To not pollute the log with warnings when an app is offline log the exception with
debug when the application status already was offline before the check. So only the
first exception when the state changes to offline will be logged.

closes #125

Since the build errors when a before_install script fails when the exitcode is non-zero we
have to use bashs regular if-then-else syntax to return a zero exitcode when $TRAVIS_TAG is not set.

Add an encrypted GnuPG keyring and use it for signing the artifacts.
When a Tag is present deploy to sonatype staging repository.