nonseparable (time+space) boundary conditions

[bmcdanie]

the current code assumes separable boundary conditions.

"unscaled" boundary condition vectors are created with some time t, and later scaled with bc_time_func(t2). I am not convinced this will yield the correct answer when the boundary condition vectors actually use the initial time t.

[hbrunie]

What is the link between the fact that BC are assumed separable, and the fact that the correct answer may not be yield when BC use initial time t?

[mkstoyanov]

The #699 interpolation framework is the right path forward for non-separable functions and operators.