refactoring ldas_setup

[weiyuan-jiang]

Initial check in. The class will be broken up into smaller steps.

This PR addresses issues #113 and #134.

Note: The ldas_setup changes impact #54 (irrigation), #131 (routing), and #132 (CatchCN51). Chances are that for each of these PRs, updating ldas_setup after the present PR has been merged will result in conflicts and require manual updates.

[gmao-rreichle]

@weiyuan-jiang : Many thanks for getting a start on this. Perhaps hold off on editing ldas_setup for refactoring until #102 and #106 have been merged. With any luck, we can merge these PRs early this week.

[gmao-rreichle]

@weiyuan-jiang : When you get back to this PR, please include an optional SLURM/PBS parameter ntasks-per-node, see issue #113. Note that unless the user specifies a value for ntasks-per-node, the default behavior should be to not specify an integer value for ntasks-per-node in the lenkf.j job file.
I don't know if it's possible to specify ntasks-per-node in lenkf.j with a blank value. If this works, I suppose we can add ntasks-per-node just like constraint. If a blank value for ntasks-per-node doesn't work in lenkf.j, we need a bit more scripting than what we do for constraint.

cc: @mathomp4

[weiyuan-jiang]

This PR addresses issue #134 and #113

[weiyuan-jiang]

@gmao-rreichle I can merge PR #132 to this PR. Should I do that ?

[weiyuan-jiang]

@saraqzhang Would you please test this PR in your GEOSadas run? I expect zero-diff. Thanks

[saraqzhang]

@weiyuan-jiang @gmao-rreichle I tested ldas_setup with ladas options. It needs one fix in setup_utils.py : add "from datetime import timedelta ". After the fix the experiments are set up correctly.

[gmao-rreichle]

@weiyuan-jiang @gmao-rreichle I tested ldas_setup with ladas options. It needs one fix in setup_utils.py : add "from datetime import timedelta ". After the fix the experiments are set up correctly.

Thanks, @saraqzhang, for taking look so quickly. @weiyuan-jiang, I added the import statement.