Add remote_read docs and update remote_write ones

Next step: add example code for remote read, point more precisely to
examples and specs for each.

Document custom query timeouts.

Next release

api: add docs for targets and alertmanagers api

Overview exporters: remove Ganglia, add Graphite

There is no Ganglia exporter listed in the [exporters doc](../instrumenting/exporters.md) and support for gmond has been deprecated from the Node exporter, so I think we should remove Ganglia from the list here. I replaced with Graphite, but open to leaving it blank or picking some other example.

Fix for overlapping text in feature item headings.

Only happens in certain window sizes.

Added /-/reload endpoint for Prometheus and Alertmanager https://github.com/prometheus/alertmanager/issues/353 (#647)

Prometheus code ref.: https://github.com/prometheus/prometheus/blob/master/web/web.go#L208

Add two exporters by Kumina

Add nginx-vts-exporter

Add exporter for Mozilla Observatory

Adding the prometheus exporter for Mozilla Observatory to docs

An extra warning is included as these functions make it easy to
produce metrics that cause duplicate time series data.

Documents prometheus#2279

Tarantool is a fast in-memory database and an application
server for Lua apps.

Homepage: https://tarantool.org

Signed-off-by: Xabier Larrakoetxea <slok69@gmail.com>

With most requests at 220ms, the true histogram should return 200-300ms
bucket for 95th percentile.

Add note on hiring technical writers

Mention Prometheus Operator in Kubernetes SD section.

Kubernetes SD: Remove references to `in_cluster`

`in_cluster` doesn't exist anymore, and the comment for the
`api_server` config option already explains how in-cluster auth works.

Add Marathon SD section to scrape and AM configs.

* Document EC2 profile configuration

* config: document Alertmanager discovery

* config: add alpha notice for Alertmanager SD