Allow for NA values in optional horizon time difference checks (#140)

[annakrystalli]

Should be reviewed after #144 & #145 are merged

This PR allows for NA values in optional horizon time difference checks by subsetting tbl to complete cases of relevant columns. This ensures rows that may not be targeting relevant to modeling task do not cause false check failure.

Resolves #140

[github-actions]

🚀 Deployed on https://67232e5437cfc011549a9895--hubvalidations-pr-previews.netlify.app

[elray1]

lgtm

[zkamvar]

LGTM!

TIL about stats::complete.cases()!

As an aside, is there a reason why the content from #144 and #145 are in here?

[annakrystalli]

As an aside, is there a reason why the content from #144 and #145 are in here?

Not sure. The diffs should have updated automatically when the other PR's were merged in. That's how it always worked before. GitHub has been a bit strange recently, like when it closed the PR when the branch it was pointing to was deleted. I'm, 100% sure in the past it just redirected the PR to main in such a situation.

[zkamvar]

Not sure. The diffs should have updated automatically when the other PR's were merged in. That's how it always worked before. GitHub has been a bit strange recently, like when it closed the PR when the branch it was pointing to was deleted. I'm, 100% sure in the past it just redirected the PR to main in such a situation.

But was this pattern you were observing based on the before R-universe workflow? In that workflow, there were always a chain of pull requests main <- A <- B <- C and they would be merged in the order of C, B, and finally A. When GitHub deleted the branch last time, it was because the same chain existed, but they were merged in the order of A, B, and finally C. When you merged A, the branch that B was pointing to no longer existed, so the pull request was closed.

In this case, you had A, B, and C all pointing to the main branch, but A, B, and C were still dependent on each other (because they were created sequentially).

I would recommend creating each new branch from main in the future.

[annakrystalli]

Sometimes I need the features in the other PRs too.

It has worked in the past and when I need to I will still use this approach

[annakrystalli]

Actually, thinking about it, it does make sense why it didn't work as I expected. As mentioned above I do often need to build on previous PRs. But what seems to be actually needed in this situation is a merge in between: isaacs/github#750 (comment) and three-dot-and-two-dot-git-diff-comparisons