Add bc that sets zero only at the incoming boundary

...even when diffusion is present. This is in contrast to "zero" which
sets zero at the incoming boundary when no diffusion is present, but
zero at both boundaries when some diffusion is present. This may be
useful when numerical diffusion is present but is small compared to the
advection speed near the boundary.

src/initial_conditions.jl

@@ -2114,6 +2114,15 @@ function enforce_v_boundary_condition_local!(f, bc, speed, v_diffusion)
             # 'upwind' boundary
             f[end] = 0.0
         end
+    elseif bc == "incoming_zero"
+        if speed[1] > 0.0
+            # 'upwind' boundary
+            f[1] = 0.0
+        end
+        if speed[end] < 0.0
+            # 'upwind' boundary
+            f[end] = 0.0
+        end
     elseif bc == "both_zero"
         f[1] = 0.0
         f[end] = 0.0

src/moment_kinetics_input.jl

@@ -1043,6 +1043,8 @@ function check_coordinate_input(coord, coord_name, io)
         println(io,">$coord_name.bc = 'constant'.  enforcing constant incoming BC in $coord_name.")
     elseif coord.bc == "zero"
         println(io,">$coord_name.bc = 'zero'.  enforcing zero incoming BC in $coord_name.")
+    elseif coord.bc == "incoming_zero"
+        println(io,">$coord_name.bc = 'incoming_zero'.  enforcing zero incoming BC in $coord_name (even when there is diffusion).")
     elseif coord.bc == "both_zero"
         println(io,">$coord_name.bc = 'both_zero'.  enforcing zero BC in $coord_name.")
     elseif coord.bc == "periodic"