Refactor the way floating wall are described

[blaisb]

The current floating wall implementation uses the following notation to specify floating walls:

subsection floating walls
set number of floating walls = 1
subsection wall 0
subsection point on wall
set x = 0
set y = 0
set z = 0
end
subsection normal vector
set nx = 0
set ny = 0
set nz = 1
end
set start time = 0
set end time = 4
end
end

My main gripe with this nomenclature is that we have moved away from using x, y , z as different variables to instead use lists of points (e.g. the point on the wall would be point on wall = 0, 0, 0) instead of using different variables. This is something that we should refactor in the short term.