chore(r): Update CI setup to silence warnings for Go-based drivers

[paleolimbot]

Closes #2707.

This is more likely to be a change in Adbc somewhere since it's showing up in all versions. Did we add anything recently that logged anything to stderr or called assert or abort?

In the event it can't be avoided (it often can't), we can update the warning fail level for go-based drivers since we don't distribute them on CRAN.

[lidavidm]

This is more likely to be a change in Adbc somewhere since it's showing up in all versions. Did we add anything recently that logged anything to stderr or called assert or abort?

I don't think we did anything recently but I wouldn't be surprised if the Go runtime calls abort in some circumstances. I wonder if we could just interpose the symbol somehow to call R instead?

[paleolimbot]

That sounds hard, but good to know there's an option! The driver packages have very little R code and so the WARNINGs aren't really catching anything anyway (the mostly likely thing that those checks are catching are things that will cause those drivers to stop building or loading, which would be ERRORs).