Multiple logstash backend relations should cause an error

Currently, if I relate filebeat:logstash to graylog:beats and filebeat:logstash to logstash:beat, the charm allows for this configuration and configures all graylog unit and logstash unit IPs in the list of logstash hosts. Unfortunately, this means only some portion of my logs end up on each backend logging platform due to the nature of the `loadbalance: true` configuration.

I think it would be useful to block status if multiple backends are related, We should only be able to relate one single application to either logstash or elasticsearch backends, as the filebeat spec only allows for a single output to be defined in filebeat.yml.

Trace how the charm handles filebeat:logstash, graylog:beats, and logstash:beat relations, then inspect how the filebeat.yml output and logstash host list are generated. Done means multiple backend relations put the charm in blocked status and a valid configuration exposes only one logstash or Elasticsearch output.

Written by the indexing model from the issue text.