Memory leak when fetching addons

[polskikiel]

Description

On our long-lasting clusters, the Helm-broker sometimes restart because of the memory/goroutines leak.

Check the graphs for the Helm Broker from the nightly cluster:

On this cluster, we are adding addons in the loop when executing the tests, the memory leak must be related to it.

Expected result

The Helm broker has a stable memory footprint.

[mszostok]

@a-thaler reported such issue:

When adding my URL to a fresh cluster: github.com/sap/xf-addons//addons/index.yaml
It looks like the broker just crashes with OOM

Next try to configure the URL result in:
{"level":"error","log":{"addon":"marketing-connector-0.1.0","message":"while loading addon: while getting Docs URL for addon marketing-connector: while creating archive '/tmp/docs-marketing-connector-0.1.0.tgz': file already exists: /tmp/docs-marketing-connector-0.1.0.tgz","time":"2020-01-08T14:32:34.012Z"}}

We need to check why this is not working as this is a blocker for using HB with git repo on faros pipelines.

Expected result:

• HB recovers after OOM or any other restarts and reprocesses failed addons