Refactor/update dots

[kylebaron]

Summary

This pull request refactors bad behavior that was in place when passing arguments to mrgsolve::update(). There was some bad reasoning behind the change introduced 4 years ago or so. It needs to be changed.

By default, there should be at least a warning issued when an argument is passed to mrgsolve::update() that the function doesn't recognize. Eventually, I think this should be an error, but historically it has been a warning so we'll keep it that way for now.

There was a (bad) option controlling this; the option was bad and the implementation was bad. So this PR deprecates the option. This is handled in the documentation where we describe package-wide options and there is a warning issued if the user passes a bad argument to mrgsolve::update() with the option set. I highly doubt anyone is using this option so I don't anticipate there will be downstream effects. But we'll warn if the option is used when the situation is right.

There is also an argument to mrgsolve::update() that controls if the warning should be issued when a bad argument is passed. This is really old, from when everything was organized quite different and we didn't have a very well defined model specification file. I think we might eventually drop that as well ... there is no need to make this behavior optional. So I wouldn't be against also deprecating the argument ... probably would be fine to just rip the band-aid off. But haven't done that (yet) in this PR.

[kyleam]

By default, there should be at least a warning issued when an argument is passed to mrgsolve::update() that the function doesn't recognize. Eventually, I think this should be an error, but historically it has been a warning so we'll keep it that way for now.

Makes sense, and the code changes look good to me.

One comment about something that's not new to this PR but related: shouldn't updates description for strict say "warning" rather than "error"?

mrgsolve/R/update.R

Lines 43 to 44 in 35ac94e

	#' @param strict if \code{TRUE}, then an error will be generated if there is
	#' attempt to update a non-existent item

It looks like it did error when that description was introduced in 33eab8f (2019-09-29) and then the description went stale in the next commit when 88c76c9 (warn when bad update is attempted from simulation function, 2019-09-29) demoted the error to a warning.

[kylebaron]

Thanks @kyleam ; i'll update that and put in a TODO to finally clean this up.