nonparametric: TestRegCoefC need checking, better tests

We should return the p-value and attach relevant variables (I added the bootstrap t_dist in my branch)

My first example reports "Not Significant" even though the parameters are large. Another run has significant at 10% without pivot, and at 1% with pivot. looks fine.
The test statistic and the bootstrap distribution have small values in my example. Should there be a normalization also in the pivot=False case?

the test only checks whether "Not Significant" is reported or not, but not any numbers

versions of the test

Racine, J. (1997): "Consistent Significance Testing for Nonparametric Regression" is not cited in the book
Book, R package np, and
Racine, J.S., J. Hart, and Q. Li (2006), “Testing the significance of categorical predictor variables
in nonparametric regression models,” Econometric Reviews, 25, 523-544.
use wild bootstrap

Racine, J.S., J. Hart, and Q. Li (2006) first bootstrap method is residual as the one implemented here.

Book has other tests in section 12.3.4, looks a bit similar to TestFForm.

• given that most pieces and the basic structure is available, it might be possible to offer more choices
  without a big amount of extra work

Start at TestRegCoefC and the existing test that checks only whether "Not Significant" is reported. Compare its behavior with the cited book and R package np references, including residual or wild bootstrap variants and pivot normalization. Done means returning the p-value and relevant variables while adding tests that check numeric results and the reported significance.