CRD source based on getting endpoints from CRD

Alibaba Cloud Provider add zone info to deployment

Add filter by service type feature

Add Alibaba Cloud Provider

* Add generic metrics for Source & Registry Errors

Change-Id: Ie145f200f815712ec372fca1c897bee3710b045b

* add label to distinguish read errors

Change-Id: I9fb1295c6fb3b2c5814ba942b4af91e1fca1c2c1

* detect write errors

Change-Id: I524360923359a5aebe6224c4e32553b138dfbdab

* revert back to label less metrics

Change-Id: Id689e0ba8c3a40a520f5af73219a4e202c167e03

* fixup Gopkg.lock

Change-Id: I3e3fdc8ad792128421b92db587bdde6e788fea58

Updating changelog for new release v0.5.5

Dyn: cache records per zone using zone's serial number

exoscale: bump version to 0.11

Set Linode user agent

Configure req timeout calling k8s APIs

Signed-off-by: Yoan Blanc <yoan.blanc@exoscale.ch>

When running in a pod sometimes the request to get ingreses/services
stalls indefinitely. A simple pod restart fixes this. Hard to reproduce
but I got lucky and did thread dump which revealed a gorouting blocked
on call to k8s.

What's new is a `--request-timeout` flag that makes requests to k8s
bounded in time. The default is 30s - this may cause some deployments
with a slow api-server to timeout.

adding assume role to aws_sd provider

Linode Provider Implementation

The only thing preventing use of a smaller interval is the API request
limit. Caching records by the zone's serial number would let users set a
smaller interval and still not hit Dyn's request limit if there aren't
any changes to the zone since the last time external-dns has run.

In a dynamic setting bigger interval is still the main throttling
mechanism.

Retry calls to dyn on ErrRateLimited

Fixes #672 Cloudflare Link Language Specificity

**Why:**

* Cloudflare link shouldn't have the langauge and should rely on the
  browser

**This change addresses the need by:**

* closes #672
Signed-off-by: Christopher Hein <me@christopherhein.com>