[Feature Request] Large number of histogram buckets

[taintedkernel]

When scraping the loadbalancing GCP API, it seems a very large amount of buckets are created for each endpoint. You can see an example in this unrelated PR https://github.com/frodenas/stackdriver_exporter/pull/23

If the number of endpoints is large, this results in an enormous number of metrics. It seems this bucket count is fixed and comes from the GCP API directly. Can these be reduced in some capacity with resampling or similar techniques?