A bunch of changes relating to HTM and Precession

1. HorizontalToEquatorial now computes J2000.0 coordinates ra0, dec0
2. Constellation boundaries should actually precess. Changing base class from NoPrecessIndex to LineListIndex and making a few changes
3. Fixed SkyMesh::intersect along a line which would give wrong results for really small lines (great circle arcs). Needs a test suite though.
4. lastPrecessJD should be long double; double precision is insufficient for JD (okay for MJD)
5. Uniformize behavior of SkyMesh::index. This presumably (not explicitly verified) fixes a bug right-clicking on stars when simulation time is far away from J2000, because that method used SkyMesh::index around JNow coordinates against a J2000.0 index. The NO_PRECESS_INDEX cases are explicitly altered by passing a useJ2000 = false argument.

It's nice to see rho Aquilae go from Aquila to Delphinus when changing the simulation time :-)

[Avoid squashing commits so they can be reverted individually if needed]