[raster] Coverage version of stats functions should be aggregates
|Reported by:||pracine||Owned by:||dustymugs|
Thinking about it twice I was wondering if it would not be better to implement the coverage versions of ST_Histogram, ST_SummaryStat, ST_count, ST_ValueCount, ST_ValuePercent (BTW should not there be a coverage version of ST_Quantile?) as aggregates. We would be able to limit the stats to tiles selected with a where clause.
I guess you would just have to refactor the existing code into a complex state object with a state function accumulating stats and a final function.