Add test.statistic and test.type argument to getAnova() inside of getCoefficients()

[qdread]

This PR addresses issue #305 but only for the specific case where all submodels are of class lmerMod.

The coefs() function includes arguments test.statistic and test.type. But currently, those two arguments are not doing anything. They are intended to be arguments to car::Anova(), passed via getAnova(), but they are not passed all the way through to getAnova(). As a result, car::Anova() never "sees" them, and changing them has no effect, even if invalid values are input.

The small correction I made here to the getCoefficients() function passes the user-specified arguments to getAnova(), but only when all(class(model) %in% c("lmerMod")) is true. Therefore it has the desired result for lmer models but does nothing otherwise. I would recommend noting that in the documentation of coefs: specifying that test.statistic and test.type arguments will only have an effect if the models are lmerMod.

Thanks for your great package and thanks for your attention to this PR!