Add documentation about PostgreSQL clustering specificity

[SCedricThomas]

In our documentation there are no information about our way to implement an HA PG cluster.
As there is several possibilities to make an HA cluster, we should document our current architecture.
It could be worth mentioning the following information if they are missing :

• We are using Patroni and etcd to achieve an election of the leader inside the cluster
• The secondary node is only available for failovers
• We are relying on physical replication only