The apt repository uses a non-existant fact to create the list, update
to a fact that does exist.

This patch fixes the Exec example adding the missing comma to provide valid syntax for the example.

Adds support for omit_hostname agent parameter

Add in an additional node definition for Ubuntu 16.04, and handle the
ability to test with Puppet 4.  To make this work, set the following
environment variables on your host:

```
PUPPET_INSTALL_TYPE=agent
PUPPET_INSTALL_VERSION=1.6.1
```

* Fix fact-related telegraf::input specs

These failures appear to have been caused by additional fact requirements
added since the specs were first written. Additionally, the my_snmp
check was failing due to a change that causes integers to be written to
the config as bare numerals, not as strings.

* Bump Travis ruby version

Required for newer versions of beaker.

* Add support for purging unmanaged config files

This currently defaults to disabled to ease migration from previous
versions.

* Use Puppet3-compatible resource syntax

This doesn't rely on File {} defaults to avoid potential scope
weirdness, however unlikely.

* Normalize Repository file for RHEL and CentOS

According to the following we could make use of `yum variables`:
https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/6/html/Deployment_Guide/sec-Using_Yum_Variables.html

Furthermore it might be preferable to set the `descr`  field to the value on the official docs instructions:
https://www.influxdata.com/package-repository-for-linux/#centos-users

* Update install.pp

* Update telegraf_spec.rb

* Update telegraf_spec.rb

copy pasterino typo :)

* Update install.pp

* Test CentOS and RHEL

* Update .travis.yml

to prevent the error message:
```
Repository telegraf is missing name in configuration, using id
```

There may be a time when a sys admin needs to stop the telegraf service
on one or more machines. Let's add the ability to _control_ the
service, rather than statically enabling and starting the service.

I'm not sure when the url changed, but the wrong url was causing yum to
error as it tried to update the influxdata repo.