Treat all documented names as known symbols

[lionel-]

Branched from #872.

Progress towards posit-dev/positron#8521
Addresses posit-dev/positron#8708

This adds all symbol-looking documentation topics to our set of known exported symbols to cover exported datasets. This information is installed in an INDEX file that we can inspect statically: https://cran.r-project.org/doc/manuals/r-devel/R-exts.html#The-INDEX-file-1. This INDEX file might be manually written by the package author (see link above), and so might throw off our attempts at parsing this unstructured file. That said I expect it's automatically generated in the vast majority of cases

This stopgap approach will potentially produce false negatives:

• Documented topics are not necessarily about exported symbols.
• This approach will indistinctly export datasets whether LazyData is true or false. In the false case, the dataset is only exported after a corresponding data() call.

For now though, we prefer to avoid false positives and spurious diagnostics.

QA Notes

In a fresh session with:

# `LazyData: true` case
penguins # Not in scope
library(palmerpenguins)
penguins

# `LazyData: false` case
liner # Not in scope
library(BCA1SG)
liner # FIXME: Should not be in scope
data(liner)
liner

You should only see diagnostics before the library() calls:

[DavisVaughan]

Reviewed together over zed