fix some bugs in cvt reconstructions

[tongtongcao]

Explanation of bugs and tests for fixes are in the slides:
Bug Fixing for CVT Reconstruction.pdf

[tongtongcao]

Opposite definition for sign of d0 between seeds and tracks causes confusion and issues.
For example, the funciton getPointAtRadius() in org.jlab.rec.cvt.trajectory.Helix is used at several places, with input of radius from a seed or a track. However, the function supposes that definition sign of d0 for seeds and tracks are the same.
With the update, definition for sign of d0 is consistent at the seeding and tracking level.
However, to keep consistent with old definition of d0 for tracks, we still change sign of d0 when writing d0 into track banks, so that we do not change codes in the following reconstruction or in analysis packages. For example, in the EB, d0 is used to set vx and vy.
I did quick check for comparison of tracks between and after the update. Difference is slight.