Use reframe() instead of summarize() for multiple rows

[capnrefsmmat]

Since dplyr 1.1.0, returning multiple rows from summarize() is deprecated, and you're supposed to use reframe() instead. These are our two use-cases, so switch them over.

I ran into this while updating an old presentation, and dplyr helpfully barfed the deprecation warning right into my slides. This should fix it.

[dajmcdon]

My memory is that the reason it's not on main was to avoid forcing @capnrefsmmat and @krivard to be required for PR review. I think at some point, there was concern that anything on main in this repo was "public" and therefore needed more stringent approvals. But some packages had (at the time) much faster dev cycles.

Longer term, I would suggest that any packages that are still useful to be moved out of subfolders in this repo into their own repos. And perhaps kill off stale junk.

[dajmcdon]

The failed evalcast test is potentially concerning. But @nmdefries is more up to speed on that package's status.

[capnrefsmmat]

My memory is that the reason it's not on main was to avoid forcing @capnrefsmmat and @krivard to be required for PR review. I think at some point, there was concern that anything on main in this repo was "public" and therefore needed more stringent approvals. But some packages had (at the time) much faster dev cycles.

Before covidcast was on CRAN, we were certainly worried that most people installed from GitHub, so changes would immediately be visible to users. Does evalcast get used by outsiders in that way, or is everyone using evalcast also involved in development?

We can also update the CODEOWNERS file if it's constantly requiring unnecessary reviews. It's probably worth reviewing it anyway, as several people have left.

[dshemetov]

Pinging @lcbrooks for evalcast usage input. I have a hunch that we're the only ones using it, but maybe not.

The current evalcast CI failure is coming from bettermc being booted off CRAN. @nmdefries made a fix already here, I'm going to apply a similar fix on the evalcast branch and merge it here.

[brookslogan]

I'm not aware of other people who are using evalcast, and I can't find any usage on GitHub (searches A, B give only false positive plus hadley's covidcast fork (probably from updating covidcast package's tidyselect usage due to CRAN checks)) or on the web. So I think we're free to do whatever we want here regarding evalcast branch management.

[Guess I should have also checked for DESCRIPTION files / etc. to see if there are packages that depend on evalcast. But the main one to check is covidHubUtils but it also doesn't use evalcast, so I'm guessing there aren't.]

[krivard]

CODEOWNERS and CI workflow look solid; minor question on working directory settings

[krivard]

👍 ci changes approved

[krivard]

for a hot second I was extremely confused, because the CI changes from this commit block are for covidcast, and this version dependency change is for evalcast

but i've confirmed that the only R version in the evalcast ci is 4.1, so this change should work fine with ci