More refactoring

[dellaert]

• added eliminate as method of HGFG to test in isolation
• moved HBN test to right place
• added three "bougie" constructors that considerably simplify tests and later examples:

/// Construct from mean `mu_i` and `sigma_i`.
  HybridGaussianConditional(
      const DiscreteKey mode, Key key,  //
      const std::vector<std::pair<Vector, double>> &parameters);

  /// Construct from conditional mean `A1 p1 + b_i` and `sigma_i`.
  HybridGaussianConditional(
      const DiscreteKey mode, Key key,  //
      const Matrix &A, Key parent,
      const std::vector<std::pair<Vector, double>> &parameters);

  /// Construct from conditional mean `A1 p1 + A2 p2 + b_i` and `sigma_i`.
  HybridGaussianConditional(
      const DiscreteKey mode, Key key,  //
      const Matrix &A1, Key parent1, const Matrix &A2, Key parent2,
      const std::vector<std::pair<Vector, double>> &parameters);

[varunagrawal]

Approve modulo comments.

I like the new API but it would also be nice to have at least one test using the old API as a tutorial, aka "this is good, this new thing is better".

[varunagrawal]

This deviates from the naming scheme of GaussianFactorGraph and other graphs, where it is frontalKeys. Should we strive for consistency?

[dellaert]

Ok, I’ll fix

[dellaert]

Actually, no,I went back and looked at GFG and it's always just "keys". So I'll leave this to be consistent. frontal keys is a concept in conditionals, not in elimination. The keys we eliminate will become frontal keys.

[varunagrawal]

Oh okay. I must have been mistaken then.

[dellaert]

I’ll address the comments. But

• I don’t want the separators between the constructors
• The includes should not depend on transitive inclusion, see Google style guide
  Maybe let me edit to avoid conflicts? You have enough to do with experiments :-)

[varunagrawal]

Sure thing wrt the dividers. It's a nitpick anyway :)

[varunagrawal]

• The includes should not depend on transitive inclusion, see Google style guide

Interesting. I like to minimize includes since that helps cut down compilation time. Less header copying.

Maybe let me edit to avoid conflicts? You have enough to do with experiments :-)

Fair. I was updating the code to examine the ordering, so decided to save a compilation cycle by making the edits.