noise models: Add getters for all model parameters

[jlblancoc]

Add missing "getters" for noise models. Otherwise, it's impossible to find out their internal parameters after construction time.

They were needed to allow the development of this new non-intrusive alternative serialization library.

[jlblancoc]

PS: I didn't add a new virtual method modelParameter() to robust Base for two reasons:

• ABI break.
• Not all robust models have 1 parameters ("Null"), and, perhaps, one day somebody creates a model with 2 or N parameters.

[dellaert]

I think this is OK and I'll approve it.
If time/resources were not an issue, I'd probably would have preferred a more "meaningful" name for each parameter, and - while we're at it -- better docs for each model (e.g., a doxygen block with the formula and a name for the parameter).
But, unless you have the time (or a student with time) I think this a good compromise.

[jlblancoc]

@dellaert I know... I also felt the same about the docs but didn't want to extend the scope of the PR too far away :-)
Take a look now, I'm much happier with the docs now, let me know your thoughts, or feel free of merging.

While documenting and going through the derivatives, I think I fixed a wrong "x 2" factor in the DCS weight, here, with the equations justifying the change here:

gtsam/gtsam/linear/LossFunctions.h

Lines 377 to 389 in 017e3cd

	/** DCS implements the Dynamic Covariance Scaling robust error model
	* from the paper Robust Map Optimization (Agarwal13icra).
	*
	* Under the special condition of the parameter c == 1.0 and not
	* forcing the output weight s <= 1.0, DCS is similar to Geman-McClure.
	*
	* This model has a scalar parameter "c" (with "units" of squared error).
	*
	* - Loss \rho(x) = (c²x² + cx⁴)/(x²+c)² (for any "x")
	* - Derivative \phi(x) = 2c²x/(x²+c)²
	* - Weight w(x) = \phi(x)/x = 2c²/(x²+c)² if x²>c, 1 otherwise
	*/
	class GTSAM_EXPORT DCS : public Base {

Please, confirm whether it's right now or not... I think the "2" should go on the second line, otherwise it would become a "4". Not a big deal probably anyhow...

[jlblancoc]

Ok, I just reverted the last "fix" of the DCS equation to make unit tests happy. If you think it was a real bug, we might open a issue ticket and handle it in a separate PR.

[dellaert]

OK, I'll merge this.
I have no time to check if it's a bug or not :-( If you are convinced, let's open an issue to record it at least.