Complete gradient tests for all CVCs

[jhenin]

TODO: check that all existing CVCs have their gradients tested by standalone functional tests (using debugGradients) or otherwise.
For vector CVCs or ones with implicit gradients, a test based on VMD scripts could be used (WIP).

Update: tentative list of CVCs that do not currently have a functional test (they are covered by regression tests).

• polar_theta
• polar_phi
• distance_xy
• aspath
• azpath
• gspath
• gzpath
• aspathCV
• azpathCV
• gspathCV
• gzpathCV
• linearCombination

Vector CVCs that will need a different testing mechanism:

• cartesian
• orientation
• distance_vec

[HanatoK]

Is it possible to implement a debugForces option similar to debugGradients? There should be a way to capture the forces from the atom groups, and compare them with the cvforce * (numerical gradients of each components of a colvarvalue).

[jhenin]

The goal of this is to include CVCs with implicit gradients in the test, right? That sounds like a good idea.

[jhenin]

Partly addressed by #838