Label QRCompactWY.T as Triangular

[jiahao]

IIRC T is upper triangular and should be labeled as such, otherwise the arbitrary memory access to the lower triangular block can be confusing.

julia> A=randn(6,6)
6x6 Array{Float64,2}:
 -1.50045    1.22618   -0.368893  -1.27154    -1.05317    -1.72158 
  0.338185   0.590917   0.845269   0.781246    0.691277    0.180603
  0.931246  -0.290888  -0.771341  -0.373031    0.051738   -0.249661
 -1.77921    0.525212  -0.438607   1.16413     0.0357546  -1.13636 
 -0.651204  -1.17231    0.421355  -0.0150839  -0.502015   -0.922738
  0.970267   1.24223   -1.28299   -0.158388   -0.144053   -0.558598

julia> qrfact(A).T
6x6 Array{Float64,2}:
 1.53849  0.592483   0.353011  -1.32743    0.162549  0.0
 0.0      1.31762   -0.73315    0.390086   0.601557  0.0
 0.0      0.0        1.50479   -0.438421  -0.116728  0.0
 0.0      0.0        0.0        1.99537   -0.141898  0.0
 0.0      0.0        0.0        0.0        1.69842   0.0
 0.0      0.0        0.0        0.0        0.0       0.0

julia> qrfact(A).T
6x6 Array{Float64,2}:
 1.53849         0.592483       0.353011      -1.32743        0.162549  0.0
 6.93603e-310    1.31762       -0.73315        0.390086       0.601557  0.0
 6.93604e-310  NaN              1.50479       -0.438421      -0.116728  0.0
 4.94066e-324    4.94066e-324   9.88131e-324   1.99537       -0.141898  0.0
 9.88131e-324    6.93604e-310   6.93603e-310   6.93604e-310   1.69842   0.0
 0.0             9.88131e-324   6.93603e-310   9.88131e-324   0.0       0.0

julia> qrfact(A).T
6x6 Array{Float64,2}:
  1.53849    0.592483    0.353011  -1.32743     0.162549  0.0
 -0.078889   1.31762    -0.73315    0.390086    0.601557  0.0
 -0.217233   0.0148274   1.50479   -0.438421   -0.116728  0.0
  0.415038  -0.0386644   0.159748   1.99537    -0.141898  0.0
  0.151907  -0.475831   -0.163803   0.04674     1.69842   0.0
 -0.226335   0.53829     0.526059  -0.0115561   0.421387  0.0