add test functions for gradient! and fisher_B!

test/two_variance_component_test.jl

@@ -52,6 +52,8 @@ vcmodelrot = TwoVarCompModelRotate(vcmodel)
 ∇ = zeros(2d^2)
 #@code_warntype gradient!(∇, vcmodelrot, vcdatarot)
 @inferred VarianceComponentModels.gradient!(∇, vcmodelrot, vcdatarot)
+@inferred VarianceComponentModels.gradient!(∇, vcmodel, vcdatarot)
+@inferred VarianceComponentModels.gradient!(∇, vcmodel, vcdata)
 @test norm(VarianceComponentModels.gradient(vcmodel, vcdata) - 
  VarianceComponentModels.gradient(vcmodelrot, vcdatarot)) ≈ 0.0
 @test norm(VarianceComponentModels.gradient(vcmodel, vcdata) - 
@@ -75,9 +77,17 @@ H = zeros(2d^2, 2d^2)
 @info "Evaluate Fisher information matrix of B"
 H = zeros(p * d, p * d)
 #@code_warntype fisher_B!(H, vcmodelrot, vcdatarot)
+# @inferred VarianceComponentVariate(Y, X, V)
+# vcdata = VarianceComponentVariate(Y, X, V)
+# vcmodel = VarianceComponentModel(vcdata)
+
+# @info "Pre-compute eigen-decomposition and rotate data"
+# vcdatarot = TwoVarCompVariateRotate(vcdata)
+# vcmodelrot = TwoVarCompModelRotate(vcmodel)
 
 @inferred fisher_B!(H, vcmodelrot, vcdatarot)
 @inferred fisher_B!(H, vcmodel, vcdatarot)
+@inferred fisher_B!(H, vcmodel, vcdata)
 @test norm(fisher_B(vcmodel, vcdata) - fisher_B(vcmodelrot, vcdatarot)) ≈ 0.0
 @test norm(fisher_B(vcmodel, vcdata) - fisher_B(vcmodel, vcdatarot)) ≈ 0.0
 @test norm(fisher_B(vcmodel, [vcdata vcdata]) -