Class constructor for family_samples

[athowes]

Could have class constructor for this.

Originally posted by @athowes in #170 (comment)

I'm a little bit unsure exactly how this would slot into the package, but we could have class constructors / validators for the family_samples classes.

Of course it would be good to check that something is a valid family_samples.

It wouldn't fix the awkwardness of e.g.

secondary_dist <- data.table(mu = 1, sigma = 1)
secondary_dist <- as_lognormal_samples(secondary_dist)
add_mean_sd(secondary_dist)

seeming a bit much (lots of code).

[seabbs]

well I think it depends on if this is likely to be the only function that operates on samples (and I think the answer is broadly no - plotting etc) and that most users won't actually use this much as they will either work with the output of epidist or some posterior predictoin functionality both of which will automatically assign this

[athowes]

I think it won't be the only function, though perhaps other functions will operate on the output of e.g. add_mean_sd say.

And I agree users won't use it much.

[athowes]

Labelling this as wontfix as I don't know that we need it.

[seabbs]

I think we no longer need this so closing.