convenience function to remove pairs with missing values

[corybrunson]

As proposed in this issue on ripserr, it may make sense to distinguish between persistent pairs with infinite values and those with missing values. The former might need to be handled in dependents like {plt} and {TDAvec} as they see fit, or else by the user; but i think it will be important for {phutil} to include a function to prune such pairs from persistence data, the reasons being that they are much less common and might be generated by one engine but not another and therefore obscured from a user's view.

This would be useful downstream, e.g. as the core of a {tdarec} step to remove pairs with missing values.

I haven't thoroughly explored the possibility space, but the most efficient method i found looks like this:

l <- lapply(l, function(m) m[! apply(is.na(m) | is.nan(m), 1L, any), , drop = FALSE])