Enable outputting data at each timestep

[spencerahill]

Even for high frequency data, the 'ts' output option is still an average over the specified sub-annual range. It would be nice to be able to save the data at every timestep on which it was created. This is a necessary condition for using aospy-generated data as input for other aospy functions, c.f. #3. The logic should be relatively straightforward to implement: compute the function at each timestep as usual, and then just skip all further time reductions.

Perhaps not right away, but ultimately there would need to be the ability to specify (or automatically determine) time chunks for splitting the output into different files. For high frequency data, e.g. 3hr or 6hr, even a 1-2 yr file is on the order of 1 GB. So without some chunking, the files will become huge for longer time durations.

[spencerahill]

#208 would cover this, so closing this as duplicate