Setting a 1 second interval for an aggregation across a long time range will generate a very large number of buckets. I suspect Kibana therefore alters the interval automatically. You should be able to check this by looking at the generated aggregation, which can be seen by expanding the statistics view in the lower left corner of the visualisation.
If you are looking for a graph showing per second request rate, I would recommend creating a Timelion sheet that uses the scale_interval function (similar to this thread). As Timelion is now part of core Kibana, you can add Timelion sheets to a dashboard.