allow for ANY as weight type backup for n_mean_var()

[malcolmbarrett]

This PR allows for n_mean_var() to take a weight of any type as long is it can pass through as.double(). Basically, it adds a c("thing", "ANY") signature for the existing input types. It also treats a conversion warning as an error.

The motivation for this PR is to allow for smd to accept weights that are classed vectors. We use it in our causal inference work and some of our tooling will be expanding in this direction

[bsaul]

@malcolmbarrett - thanks for the PR. Happy to discuss the proposal. I hesitate to add support for dispatching on "ANY" class of weights, as I'm generally averse to weakening what limited form of type-like safety we can get in R (I recognize you're trying to catch ill-defined behavior). I'll think on other ways to get what you're after.

Also, if you want to the w argument to dispatch on "ANY" why not the x argument too?

[malcolmbarrett]

You can see the type of class I hope to extend this to here r-causal/propensity#10, if it helps give you any ideas. I don't use S4, so maybe there is a better way. But I don't see this being much of an issue in practice because as.double() serves as a pretty good guard (except I wish it would error rather than warn in many cases!)

Anyway, thanks for considering. It would help us a lot to be able to be more flexible here

[malcolmbarrett]

Also, if you want to the w argument to dispatch on "ANY" why not the x argument too?

w is better defined in general than x. It needs to be a numeric or castable to a numeric. But I don't have a strong opinion about it since x works well as-is for our needs

[malcolmbarrett]

This is a very small perspective into the stability of this change as I think the bigger concern is to direct users, not developers, but reverse dependencies do well with this PR

> revdepcheck::revdep_check()
── INIT ───────────────────────────────────────────── Computing revdeps ──
── INSTALL ───────────────────────────────────────────────── 2 versions ──
Installing CRAN version of smd
Installing DEV version of smd
── CHECK ─────────────────────────────────────────────────── 4 packages ──
✔ cardx 0.2.2                            ── E: 0     | W: 0     | N: 0    
✔ gtsummary 2.0.4                        ── E: 0     | W: 0     | N: 0    
✔ smdi 0.3.1                             ── E: 1     | W: 0     | N: 0    
✔ tidysmd 0.2.0                          ── E: 0     | W: 0     | N: 0    
OK: 4                                                                   
BROKEN: 0
Total time: 14 min

Of course, your main reverse import is from me 😝

[bsaul]

@malcolmbarrett - sorry the delay. This seems good. Can you:

• bump the minor version in DESCRIPTION
• optional but encouraged: add yourself as a contributor in DESCRIPTION

Thanks for the contribution!

[malcolmbarrett]

Done, thanks!

[bsaul]

Thanks!

[bsaul]

@malcolmbarrett - I'll submit to CRAN in the next few days.

[malcolmbarrett]

Much appreciated, @bsaul! No hurry on our end

[bsaul]

FYI - submitted to CRAN. Usually takes a couple of days to show up.

[bsaul]

Dear maintainer,
 
thanks, package smd_0.8.0.tar.gz is on its way to CRAN.
 
Best regards,
CRAN teams' auto-check service
Package check result: OK

No changes to worse in reverse depends.