Support native CLHEP unit system

[sethrj]

This allows users to configure Celeritas to process "native" values in a different unit system. Having multiple unit systems supported by the code in CI provides a valuable check for correctness of units in physics code, unit tests, and translation layers.

Following on to #1076 as a holiday/vacation project, this fixes several unit system scaling issues in Celeritas and fixes (or disables) issues with missing system of units in the unit tests.

• Add missing scaling units to event writer
• Replace hardcoded CLHEP::cm with a variable clhep_length which translates from CLHEP to the native unit system
• Add units to ImportData so that we can convert in place after importing
• Add runtime enumeration for doing unit conversions on import quantities
• Activate VecGeom unit system scaling routines

Things to note:

• All data is converted to native unit system as soon as possible, and it's always exported in the native unit system
• VecGeom hardcodes "bump" values as fixed-precision values (rather than specific to a length scale), so we cannot get exactly the same results when using different scale factors.
• CGS doesn't work for JSON ORANGE problems at present because the scales are embedded into the preprocessed data (this will obviously change when we create geometries on the fly)
• Single-precision runtime arithmetic messes with the imported values because the unit system, like the constants, are stored as "real_type" rather than double. (This is to avoid accidentally introducing floating point arithmetic at runtime).

Changes split into separate PRs:

• vecgeom link fix is in Fix link errors in edge cases and restore HIP testing #1042
• unit traits are in Define unit traits and support arbitrary RZ field map units #1088
• ROOT include fix in Fix and avoid including algorithms header for ROOT/CLING #1087

[stognini]

Thanks Seth! Looks great.

[amandalund]

Cool, this is great @sethrj!

Are the tests expected to pass in all unit systems now? I'm still seeing a handful failing with SI and CLHEP (looks like device tests).

[amandalund]

It looks like you already updated part of this test... were you planning on updating the rest?

[sethrj]

I got lazy at some point... I just wanted enough to show that the code essentially works, and later (next Christmas?) polish everything to a shine.

[amandalund]

Tests are passing now with CLHEP, but I'm still seeing the following failures with SI:

	 88 - celeritas/Constants (Failed)
	 96 - celeritas/em/LivermorePE (Failed)
	103 - celeritas/em/UrbanMsc (Failed)
	111 - celeritas/ext/GeantGeo (Failed)
	113 - celeritas/ext/GeantImporter:FourSteelSlabs* (Failed)
	115 - celeritas/ext/GeantImporter:OneSteelSphere.* (Failed)
	124 - celeritas/field/LinearPropagator (Failed)
	136 - celeritas/global/AlongStep:SimpleCmsRZFieldAlongStepTest.msc_rzfield (Failed)
	166 - celeritas/phys/Physics (Failed)
	186 - celeritas/track/TrackSort (Failed)
	208 - app/demo-rasterizer (Failed)

[sethrj]

Ah I never even tested the whole thing with SI (only tested the units/constants). How about I change the title and fix SI in a subsequent PR? :)

[amandalund]

Haha, sure that works. Thanks!