Removes diffusivity calculation from DiffusivityFields() for Smagorinsky closure

[tomchor]

This PR removes the calculation of the diffusivities in DiffusivityFields() for the SmagorinskyLilly so that binary operations are avoided.

Closes #2869

[glwagner]

Perfect, does this also solve the issue?

[tomchor]

Perfect, does this also solve the issue?

Yes! I tested it with up to 10 tracers:

┌ Info: NonhydrostaticModel{GPU, RectilinearGrid}(time = 0 seconds, iteration = 0)
│ ├── grid: 8×8×6 RectilinearGrid{Float64, Periodic, Periodic, Bounded} on GPU with 3×3×3 halo
│ ├── timestepper: QuasiAdamsBashforth2TimeStepper
│ ├── tracers: (b, τ1, τ2, t3, t4, t5, t6, t8, t9, t10)
│ ├── closure: SmagorinskyLilly: C=0.1, Cb=1.0, Pr=(b = 1.0, τ1 = 1.0, τ2 = 1.0, t3 = 1.0, t4 = 1.0, t5 = 1.0, t6 = 1.0, t8 = 1.0, t9 = 1.0, t10 = 1.0)
│ ├── buoyancy: Nothing
└ └── coriolis: Nothing
[ Info: Initializing simulation...
[ Info:     ... simulation initialization complete (79.486 ms)
[ Info: Executing initial time step...
[ Info:     ... initial time step complete (4.737 minutes).
[ Info: Simulation is stopping after running for 4.739 minutes.
[ Info: Model iteration 10 equals or exceeds stop iteration 10.