Inventory default behavior

[matthewhanson]

from rob:
Can we make a date range and product name required for inventory?
That will force the user to at least think about what they are doing.
Currently if you type "modis inventory", you will wait longer than I have ever been willing to wait.

[matthewhanson]

Providing inventory a product name doesn't do anything in most cases. It's only for those datasets that have multiple assets. However to further complicate things products each have their own command line switch which means logic would have to be implemented that each one product is specified (this really should be done as an ArgumentParser subclass). Given these two things I'd say it doesn't make sense to make product required.
Date range could be required, however I frequently use inventory without dates to see the span of all data for a given region. I don't think specifying a date range makes sense.
However, I think in every case, no matter the data, a spatial extent should be specified. I can't think of a reason, other than testing, where someone would want to get inventory for the whole globe. Since spatial extent can be specified via --site or --tiles this isn't as easy as adding a required=True keyword. Argparse does support mutually exclusive groups where you require one of the keywords, however there are use cases (RMA is one) where both --site and --tile are specified (e.g. to warp one specific tile). So what is needed is a check that at least one of those switches is present. I'll look into it.

[mcorbiere]

Also, gips_process Landsat will try to work on every scene in the archive.

[ircwaves]

So, perhaps we could add an option --complete-inventory (or the like) that will enable that usecase without resulting in undesirable behavior of sitting forever while inventorying everything.