[Centralized config mgmt] Error message when .logstash does not exist

*Original comment by @ycombinator:*

### Steps to reproduce
1. Run Elasticsearch, Logstash, and Kibana with x-pack.
2. Either start with a fresh install of all the above products or `DELETE .logstash` to simulate a fresh install.
3. Configure Centralized Configuration Management in Logstash:

```
xpack.management.enabled: true

xpack.management.elasticsearch.username: elastic
xpack.management.elasticsearch.password: XXXXXXXX
xpack.management.elasticsearch.url: "http://localhost:9200"
xpack.management.pipeline.id: [ "apache", "test" ]
```
4. Start up Logstash: `./bin/logstash`

### Observed behavior

The Logstash log shows errors like this every 5 seconds (or whatever `xpack.management.logstash.poll_interval` is set to in `logstash.yml`):

```
[ERROR] 2017-09-26 07:44:53.530 [Ruby-0-Thread-2: /Users/shaunak/logstash-6.0.0-rc1/vendor/bundle/jruby/2.3.0/gems/stud-0.0.23/lib/stud/task.rb:22] sourceloader - Could not fetch all the sources {:exception=>LogStash::ConfigManagement::ElasticsearchSource::RemoteConfigError, :message=>"Empty configuration for pipeline_id: apache", :backtrace=>["/Users/shaunak/logstash-6.0.0-rc1/vendor/bundle/jruby/2.3.0/gems/x-pack-6.0.0.rc1-java/lib/config_management/elasticsearch_source.rb:96:in `get_pipeline'", "/Users/shaunak/logstash-6.0.0-rc1/vendor/bundle/jruby/2.3.0/gems/x-pack-6.0.0.rc1-java/lib/config_management/elasticsearch_source.rb:82:in `block in pipeline_configs'", "org/jruby/RubyArray.java:2481:in `collect'", "/Users/shaunak/logstash-6.0.0-rc1/vendor/bundle/jruby/2.3.0/gems/x-pack-6.0.0.rc1-java/lib/config_management/elasticsearch_source.rb:81:in `pipeline_configs'", "/Users/shaunak/logstash-6.0.0-rc1/logstash-core/lib/logstash/config/source_loader.rb:59:in `block in fetch'", "org/jruby/RubyArray.java:2481:in `collect'", "/Users/shaunak/logstash-6.0.0-rc1/logstash-core/lib/logstash/config/source_loader.rb:58:in `fetch'", "/Users/shaunak/logstash-6.0.0-rc1/logstash-core/lib/logstash/agent.rb:148:in `converge_state_and_update'", "/Users/shaunak/logstash-6.0.0-rc1/logstash-core/lib/logstash/agent.rb:105:in `block in execute'", "/Users/shaunak/logstash-6.0.0-rc1/vendor/bundle/jruby/2.3.0/gems/stud-0.0.23/lib/stud/interval.rb:18:in `interval'", "/Users/shaunak/logstash-6.0.0-rc1/logstash-core/lib/logstash/agent.rb:94:in `execute'", "/Users/shaunak/logstash-6.0.0-rc1/logstash-core/lib/logstash/runner.rb:362:in `block in execute'", "/Users/shaunak/logstash-6.0.0-rc1/vendor/bundle/jruby/2.3.0/gems/stud-0.0.23/lib/stud/task.rb:24:in `block in initialize'"]}
```

This seems to be happening because the `.logstash` index does not exist yet. For testing purposes, if you manually create the `.logstash` index (`PUT .logstash`) these errors will go away. Normally the `.logstash` index will be created by the Centralized Config Management Kibana UI when the first pipeline is created via that UI.

### Expected behavior
I would think we should not show this error when `.logstash` does not exist. That would then match the behavior when the `.logstash` index does exist but no pipeline is configured in it yet. Currently the error is not actionable and just ends up adding noise to the logs.

Reproduce the missing-.logstash case with the Centralized Configuration Management settings shown, then inspect config_management/elasticsearch_source.rb around get_pipeline and pipeline_configs. Compare the logs with an existing empty .logstash index and trace the fetch through logstash-core/lib/logstash/config/source_loader.rb. Done means the missing index no longer emits a repeated, non-actionable error.

Written by the indexing model from the issue text.