Storage Access API - retrieve data from minimal build

[simonaoliver]

Use results returned by the Search API(#13) and provide:

• construct analysis array elements from storage units
• labeled data (with xray)
• lazy loading to facilitate out of core processing (with dask for example)
• group data by inter-operability
  • same projection
  • same coordinate extents
  • same 'resolution'

[petewa]

Is this the replacement for gdf.get_data?

If so, get_data needs to be have to take as input:
- depending on how the data is stored in the storage units, a target projection may need to be specified in the get_data call to return the data in the appropriate projection.
- grouping by time. e.g. seasons, month, day etc like in the gdf trial get_data code with grouping_function
- dimension ranges or polygons or dimension masks
- target resolution.
- multi-scale version of get_data (not for v2 but should think about this, it's something I will be incorporating later on)

Arrays needs to be wrapped as a xray dataarray and metadata set as appropriate.

Reads by multi-process reads not multi-threaded reads, useful for scaling up on single node multi-core reads.

We need to have a chat about this, perhaps face to face with a whiteboard or over Skype?

[petewa]

I have created a page to discuss the Data Access Interface here: http://www.datacube.org.au/wiki/Data_Access_Interface

Here is the style guide for the wiki: http://www.datacube.org.au/wiki/Style_Guide

[v0lat1le]

@petewa Yes, it is the replacement for get_data. Look at the band_stats_integrated.py in the examples directory.