Fix column integrals for deep atmosphere

[dennisYatunin]

This PR modifies the column_integral_definite! and column_integral_indefinite! functions for deep atmosphere configurations by weighting the discrete lengths Δz with discrete areas ΔA. Since the values of ΔA are constant within each column when using the shallow atmosphere approximation, this change has no effect for shallow atmosphere configurations, and all current unit tests generate the same results.

The purpose of this change is to fix the errors observed in the deep atmosphere PR's buildkite run, where the vertical integrals for zero-moment microphysics appear to not be conserving water mass.

• Code follows the style guidelines OR N/A.
• Unit tests are included OR N/A.
• Code is exercised in an integration test OR N/A.
• Documentation has been added/updated OR N/A.

[charleskawczynski]

Suggested change

	area differential `J(z)/Δz`, with `J(z)` denoting the metric Jacobian. The input
	area differential `J(z)/Δz(z)`, with `J(z)` denoting the metric Jacobian. The input

both are a function of z, correct?

[dennisYatunin]

Yes, they are both functions of z. But there were too many zs in this docstring so I've since removed them to improve readability.

[charleskawczynski]

Nice, thank you, @dennisYatunin!

Can we please add deep atmosphere unit tests in ClimaCore for this function?

[dennisYatunin]

@charleskawczynski I have added a unit test to test/Operators/integrals.jl that checks for integral consistency in both shallow and deep configurations. We are now always able to achieve consistency up to several times machine epsilon, including with topography. The consistency is also exact for deep atmospheres with Float64 values, which is even better than I was expecting.

I had to add some keyword arguments to the space constructors in order to ensure that topography and deep atmospheres are initialized correctly. Please take another look when you have some time.