Use rocklets in DESCRIPTION and update GHA workflow accordingly

[GegznaV]

if you use devtools::document() without specifying roclets, it will use the roclets specified in the DESCRIPTION
So should we update all out DESCRIPTION files to include rocklets and then update GHA workflow to use

    - name: Roxygenize
      run: |
        devtools::document()
      shell: Rscript {0}

instead of

      - name: Roxygenize
        run: |
          devtools::document(roclets = c('rd', 'collate', 'namespace'))
          devtools::document(roclets = c('vignette'))
        shell: Rscript {0}

?

Bryan Hanson
I like this approach as among other things it makes the workflows simpler.

Vilmantas Gegzna
We have to test it with hyperSpec as here are 2 rows for rocklets. There was some kind of issues that led to having devtools::document() used 2 times.

Claudia Beleites

I was about to try it with hySpc.read.spe on your feature branch

Claudia Beleites
(then some phone calls came in...)

Vilmantas Gegzna
Is it possible to get the rocklets programmatically?
As in some workflows, we shouldn't build vignettes as this is an unnecessary step.

[cbeleites]

Is it possible to get the rocklets programmatically?

We could use roxygen2::load_options()$roclets

[cbeleites]

I adapted and tested the necessary changes in r-hyperspec/hySpc.read.spe#35 (i.e. locally for hySpc.read.spe)

@GegznaV and @bryanhanson please have a look - I think both changes should be safe to include into the skeleton regardless of using roxut or not.
Packages with vignette would need to specify roclets including "vignette" in their DESPRIPTION, though. (roxygen2::load_options()$roclets returns c("collate", "namespace", "rd") if no roclets are specified)

[GegznaV]

The only place where possible issues may arise is hyperSpec package. In hySpc.read.spe, there will be no problems. Thanks @cbeleites