Better ensemble/SA filenames

[rykelly]

Both ensemble and sensitivity analysis now save files with ensemble id in the file name. This prevents previous analyses from being automatically overwritten. In particular, when iteratively updating an analysis it is useful to be able to see how model output is changing.

This didn't seem to affect much of the code, but I grepped for and fixed the few places where these files need to be accessed by name.

Currently, the ensemble id needed to name/access files gets passed around in the settings list. This isn't ideal, since it gets overwritten any time a new analysis is performed. Thus PEcAn will always look for the most recent files, which is OK, but better might be if we kept a record of all the ensemble IDs generated within the workflow. I'll work on that in the future, but for now this is a good step.

Also included a separate commit that allows uniform parameter sampling to actually be used when generating the ensemble. The functionality was in there, but with no way to pass an argument to ask for it when using the usual workflow.

[mdietze]

Nice. While this is still fresh in your head, do you think you could update all the variable-specific files to include that variable in the file name. Likewise, all time period specific files should include the time period in the filename as well. If you look at ensemble.analysis.*.pdf you'll see an example of where I included both in the file name. Some files include all variables (e.g. the sensitivity analysis), while others are variable specific (e.g. variance decomposition), but I'm pretty sure all data extractions and figures are the average over a specific time period (which one might want to vary).