Move overland roughness assignment out of time cycling disaggregation routine

[aubreyd]

TYPE: enhancement

KEYWORDS: overland, parameters

SOURCE: Aubrey D, NCAR

DESCRIPTION OF CHANGES: The overland roughness parameter is being setup on the high-res grid in the disaggregation routine, but this routine is called every LSM timestep and the overland roughness parameter is not time-varying. It should be moved into the initialization routine.

ISSUE: The GitHub Issue that this PR addresses. For issue number 123, it would be:

Fixes #123

TESTS CONDUCTED: Gives identical solution to pre-change on 1 core. DOES NOT CURRENTLY PASS NCORES... (help!).

Checklist

• [ ] Closes issue #xxxx (An issue must exist or be created to be closed. The issue describes and documents the problem and general solution, the PR describes the technical details of the solution.)
• [ ] Tests added (unit tests and/or regression/integration tests) NOT NEEDED.
• [x] Backwards compatible
• [ ] Requires new files? If so, how to generate them. NO NEW FILES REQUIRED.
• [ ] Fully documented. INTERNAL CODE PERFORMANCE CHANGE ONLY - NO NEW FEATURES.
• [ ] Short description in the Development section of NEWS.md NOT NEEDED.

[scrasmussen]

Just wanted to say that I've been able to reproduce that one core gives an identical solution and more than one core doesn't. I'm currently looking into the issue and will try to get it fixed as soon as possible! (First real deep dive into the code base so taking longer than expected)