Startup error when amending default Chronograf retention policy

After amending the retention policy for the Chronograf database in InfluxDB[*] as per [these instructions](https://www.influxdata.com/blog/tldr-influxdb-tech-tips-june-01-2017/), the following error is printed in the kapacitor.log file each time the service starts:

```
ts=2018-08-10T16:00:35.892+01:00 lvl=error msg="failed to create database" service=kapacitor task_master=main task=chronograf-v1-7df24795-a787-4881-87d4-4bd62750917e node=influxdb_out5 err="retention policy conflicts with an existing policy" database=chronograf cluster=
```

The retention policy now looks like the following (28d):

```
InfluxDB shell version: 1.6.1
> SHOW RETENTION POLICIES ON chronograf
name duration shardGroupDuration replicaN default
---- -------- ------------------ -------- -------
autogen 672h0m0s 168h0m0s 1 true
```

[*] Because I don't want alerts hanging around forever :wink:

Start by reproducing the startup with the Chronograf database retention policy set to 28 days, then inspect the reported kapacitor.log error and the InfluxDB retention-policy state shown in the issue. Determine why startup attempts to create a conflicting policy and verify that the service starts without the error while retaining the intended policy.

Written by the indexing model from the issue text.