Update documentation of lsmeans

[gowerc]

Closes #412

For reference the code changes are just:

• Re-wrote lsmeans documentation
• Added an additional unit test with cont*cont interactions to ensure that is working as expected
• Mild re-work of the .weights="equal" implementation as my original code was very hard to read (no change in what it produces though)
• Added an @export tag to an S3 method (this was technical a bug could have caused some niche issues due to how R looks up s3 methods)

[github-actions]

CLA Assistant Lite bot ✅ All contributors have signed the CLA

[gowerc]

I have read the CLA Document and I hereby sign the CLA

[nociale]

That's all right, I made just few comments about the documentation

[gowerc]

@nociale ah, after sifting deeping into the emmeans documentation it looks like they do also support the counter factual approach with:

iris2 <- iris[ iris$Species %in% c("versicolor", "virginica"), ]
iris2$Species <- factor(iris2$Species)

mod <- lm(Sepal.Length ~ Species + Petal.Length*Petal.Width, data = iris2)

emmeans(
    mod,
    specs = "Species",
    counterfactuals = "Species"
)

Will add this to our unit tests to ensure we match them

[gowerc]

@nociale, I updated the documentation and code according to our g-chat. I closed all the prior comments because its been reworked / rewritten a fair bit since your first review apologies. Would you mind re-reviewing now please.

I also decided to delete all mentions of "weighting" in the ancova / ancova_single functions and instead just inherit the weighting description directly from lsmeans to ensure that all of our documentation regarding this is consistent.

[nociale]

Thank you Craig!!