Set the correct root level on restart

[lroberts36]

PR Summary

Currently, on restart Mesh::root_level and Mesh::current_level get set to levels relative to the legacy tree, which is incorrect (see issue #1051). This PR fixes the issue.

PR Checklist

• Code passes cpplint
• Code is formatted
• Changes are summarized in CHANGELOG.md
• CI has been triggered on Darwin for performance regression tests.
• (@lanl.gov employees) Update copyright on changed files

[Yurlungur]

LGTM

[pgrete]

Alright, this explains some of the issues I've been seeing.

With current develop I got

Number of physical refinement levels = 2
Number of logical  refinement levels = 4

for a sample restart from the input file of the advection example.
With this branch I get the

Number of physical refinement levels = 2
Number of logical  refinement levels = 2

that matches the original run.

So just double checking: The GetLegacyRootLevel call for the output is correct, isn't it?

[pgrete]

Also what worries me is that we didn't catch this with the regression tests (particularly as current develop segfaults on restart with the advection example...).
Did this just pass by chance?

[lroberts36]

@pgrete: GetLegacyRootLevel for output is correct for as long as we are outputting legacy locations. This PR just ignores that value on restart and gets the new root level from the forest when it is rebuilt.

I am a little surprised develop segfaulted for restarts. When I was working on the forest PR, I tested restarting sparse advection with non-trivial forests and it seemed to work. That being said, I agree we might need to make the restart test more robust.