Pre-filter data for date_histogram aggregations

For 1.5, what is the preferred way to pre-filter the documents that a date_histogram aggregation operates on?
For example if I want to create 1-minute buckets only over documents with timestamp now-3615s to now-15s?

Is a nested aggregation with as the outer value and date_histogram as the inner aggregation the way to go?

Also, if filter aggregation is the way to go, and I want to use several date_histogram aggregations with different extended_bounds, but the same interval length, are there any issues with nesting them all under the same filter aggregation?

Performance-wise, when using date-partitioned indices, based a timestamp field, can timestamp range filters or filter aggregations use some internal optimization to automatically determine which indices do not need to be considered for a query, or is it better to pre-select the indices by date which you know upfront contain the selected time range?