Back to first order LAGRANGE for 2d-spot.

- Still fails around 2e-4 seconds with element inversion. Curvature
  contribution is completely zero which I DO believe because the face is a
  linear edge with no curvature. Whereas for 3D calculations, the face is
  a 2D bilinear QUAD4 which can indeed have some curvature. This is why I
  see non-zero curvature in 3D calculations in e.g. kc_jacobian_tester

test/tests/kc/2d-one-spot.i

@@ -25,8 +25,8 @@ pooldepth=.2e-3
   nx = 7
   ny = 2
   displacements = 'disp_x disp_y'
-  elem_type = QUAD9
-  uniform_refine = 2
+  # elem_type = QUAD9
+  uniform_refine = 5
 []
 
 [Problem]
@@ -36,7 +36,7 @@ pooldepth=.2e-3
 [Variables]
   [./vel_x]
     scaling = 1e3
-    order = SECOND
+    # order = SECOND
     [./InitialCondition]
       type = ConstantIC
       value = 1e-15
@@ -45,7 +45,7 @@ pooldepth=.2e-3
 
   [./vel_y]
     scaling = 1e3
-    order = SECOND
+    # order = SECOND
     [./InitialCondition]
       type = ConstantIC
       value = 1e-15
@@ -54,20 +54,20 @@ pooldepth=.2e-3
 
   [./T]
     scaling = 1e-4
-    order = SECOND
+    # order = SECOND
   [../]
 
   [./p]
     scaling = 1e8
-    order = SECOND
+    # order = SECOND
   [../]
   [./disp_x]
     scaling = 1e6
-    order = SECOND
+    # order = SECOND
   [../]
   [./disp_y]
     scaling = 1e6
-    order = SECOND
+    # order = SECOND
   [../]
 []
 
@@ -306,7 +306,7 @@ pooldepth=.2e-3
 
 [Preconditioning]
   [./SMP]
-    type = FDP
+    type = SMP
     full = true
     solve_type = 'NEWTON'
   [../]
@@ -354,7 +354,7 @@ pooldepth=.2e-3
 
 [Adaptivity]
   marker = combo
-  max_h_level = 3
+  max_h_level = 5
 
   [./Indicators]
     [./error_x]