Add comparison function for dataframes which can handle both isapprox and isequal column types

[schlichtanders]

The case is for building tests around DataFrames.jl

(At least) two rather old discourse issue have been opened around this

• mentioning isapprox with recent comments (because of myself)
• mentioning DeepDiffs.jl

I couldn't find a satisfying solution despite these discourses are pretty old. Even if the implementation is not going to be inside DataFrames.jl finally, I think it is a good place to start an issue here about it.

EDIT: Here an example for rich comparison, taken from the discourse thread.

I would summarize the missing feature as follows:

• have a @test_compare_dataframes macro or similar which can truly deal with any kind of types (especially those who need isequal for comparison and those who need isapprox). It would be super helpful if such a test macro could have rich output explaining the mismatch in detail if a mismatch occured (similar to DeepDiffs)
• also make the underlying method available outside the @test-framework for who knows what

[schlichtanders]

I got a hint that the TestingUtilities package also has a some support for comparing DataFrames. I haven't tested it.

[bkamins]

I think that the approach to this should be:

1. Define the requirements.
2. Have a reference implementation.
3. Then decide if it should be shipped with DataFrames.jl or some other package (e.g. it might turn out that proper styling would require extra dependencies, which we might want to avoid)