Re-organize PEcAn packages to have more precise purposes

[ashiklom]

I think that we currently have too few packages that each do too many things. So I wanted to start a discussion about specific functionality from certain packages that can and should be isolated. Here are some initial ideas:

• Put all logger.* functions to a pecan.logger package
• Create a small set of date-related functions, and put them in a pecan.dates package that can be used by met.process and other date-related packages.
• Pull functions related to MCMC analysis (e.g. convergence checking, burnin calculation, distributions) out of the rtm, assim.batch, and utils packages (and wherever else they are hiding) into a pecan.mcmctools package.

From Hadley's package recommendations...

I believe that packages that have a wide audience should strive to do one thing and do it well. All functions in a package should be related to a single problem (or a set of closely related problems).

From the UNIX philosophy:

Write programs that do one thing and do it well.

[mdietze]

Yeah, I agree that utils in particular has gotten broad and unwieldy -- I'm open to someone going through that bit in particular post-AGU. I think most of the others aren't bad in terms of focus.

[mdietze]

A related thing I've been thinking about is a slight re-organization of the packages in the root directory (utils, db, vis, etc) -- I think there should be a folder that's explicitly for 'core' tools so that all packages are in one of core, modules, and models. I think it might make it easier for new users to follow our conceptual organization

[ashiklom]

Closing this because I think we've actually done a decent job of this.