LANLGeoMag version 2.0.0-alpha

[drsteve]

Branch for finalizing LGM v2. The sets of changes currently across other branches or PRs/draft PRs will be slowly migrated to this branch, with any updates or cleanup required along the way.

Once complete and tested, this branch will be merged into master and tagged as version 2.0.0-alpha.

Changes:

• Bug fix: Correct altitude for lambda integrand in L* calculation
  • Update regression test to reflect expected change
• Bug fix: Correct angle in Lgm_EarthEclipse
• Bug fix: Correct intersection calculation for Lgm_Ellipse
• API change: Add RelaxTolerance argument to FindShellLine
• API change: Add ErrorStatus to all function variants of ComputeI_FromMltMlat
• New function: Lgm_Terminator now added to library
• Added trap in TS04 in case of evaluation deep inside Earth
• Initial versions of Gaussian Process Regression routines
• Updates for 2nd order calculation of 1st invariant
• Extra checks in Lgm_QinDenton to reduce unnecessary output and check for inputs with standard QD naming convention ([1min, 5min, hour], in that priority order)
• Add FluxTubeVolume and Lgm_ReplaceSubString2
• Some updates to example codes (e.g., new Geodetic2)
• Tighten default magstep tolerance (Bulirsch-Stoer relative tol from 1e-5 to 0; abs tolerance remains unchanged at 1e-5)
• Added Lgm_B_FromScatteredData6
• Added preliminary functions for conversion to magnetic field-aligned coordinates (PQB)

[drsteve]

There are still a handful of changes from #39 that aren't reflected here. These fall into a few key categories:

• KdTree updates: These should be easy enough to pull over, but need some more untangling of commits
• FluxToPsd updates: These are API-breaking and the new functionality needs pulling out into new functions. This will be a separate branch/PR.
• Shabansky orbit handling updates: This currently breaks the tests. If there are pieces I can pull in here without too much effort then I will. Otherwise this is a separate branch/PR to finish including the updates.
• Additional minor changes that need testing before merge (e.g., Maxwell-Juttner updates, Lgm_Trace updates for minimum finding)

I'm taking this out of draft status as a sign of near-completion. Remaining updates will be triaged and either moved to a separated effort to incorporate or included here prior to merge. This branch is just about ready for testing with any downstream applications.

[drsteve]

To be done post merge:

• Tag as v2.0.0-alpha so we get a Zenodo release and a GH tag. Finalize as v2.0.0 after testing with downstream apps
• New PR to fix up GPRv3 and include it in the build
• New PR to update the initialization/setter/destructor for MagEphemData and MagEphemInfo for consistency with structures
• New issue to log that a way of setting the PN model is required for calculating the eclipse flag in WriteMagEphemData
• Review whether a4481a5 (in Feature expansion and some cleanup #39) can/should be merged. Currently it seems not, due to missing text files.
• New PR to resolve Lgm_Trace updates
• New PR to resolve Shabansky handling updates
• New PR to resolve updates adding uncertainty tracking to FluxToPsd, etc.