[SPARK-32497][INFRA] Installs qpdf package for CRAN check in GitHub A…

…ctions

### What changes were proposed in this pull request?

CRAN check fails due to the size of the generated PDF docs as below:

```
...
 WARNING
‘qpdf’ is needed for checks on size reduction of PDFs
...
Status: 1 WARNING, 1 NOTE
See
  ‘/home/runner/work/spark/spark/R/SparkR.Rcheck/00check.log’
for details.
```

This PR proposes to install `qpdf` in GitHub Actions.

Note that I cannot reproduce in my local with the same R version so I am not documenting it for now.

Also, while I am here, I piggyback to install SparkR when the module includes `sparkr`. it is rather a followup of SPARK-32491.

### Why are the changes needed?

To fix SparkR CRAN check failure.

### Does this PR introduce _any_ user-facing change?

No, dev-only.

### How was this patch tested?

GitHub Actions will test it out.

Closes apache#29306 from HyukjinKwon/SPARK-32497.

Authored-by: HyukjinKwon <[email protected]>
Signed-off-by: HyukjinKwon <[email protected]>

.github/workflows/master.yml

@@ -157,7 +157,8 @@ jobs:
     - name: Install R packages
       if: contains(matrix.modules, 'sparkr')
       run: |
-        sudo apt-get install -y libcurl4-openssl-dev
+        # qpdf is required to reduce the size of PDFs to make CRAN check pass. See SPARK-32497.
+        sudo apt-get install -y libcurl4-openssl-dev qpdf
         sudo Rscript -e "install.packages(c('knitr', 'rmarkdown', 'testthat', 'devtools', 'e1071', 'survival', 'arrow', 'roxygen2'), repos='https://cloud.r-project.org/')"
         # Show installed packages in R.
         sudo Rscript -e 'pkg_list <- as.data.frame(installed.packages()[, c(1,3:4)]); pkg_list[is.na(pkg_list$Priority), 1:2, drop = FALSE]'

dev/run-tests.py

@@ -557,8 +557,14 @@ def main():
 
     # Install SparkR
     should_only_test_modules = opts.modules is not None
-    if not should_only_test_modules:
+    test_modules = []
+    if should_only_test_modules:
+        str_test_modules = [m.strip() for m in opts.modules.split(",")]
+        test_modules = [m for m in modules.all_modules if m.name in str_test_modules]
+
+    if not should_only_test_modules or modules.sparkr in test_modules:
         # If tests modules are specified, we will not run R linter.
+        # SparkR needs the manual SparkR installation.
         if which("R"):
             run_cmd([os.path.join(SPARK_HOME, "R", "install-dev.sh")])
         else:
@@ -584,15 +590,11 @@ def main():
     print("[info] Using build tool", build_tool, "with Hadoop profile", hadoop_version,
           "under environment", test_env)
 
-    changed_modules = None
-    test_modules = None
-    changed_files = None
+    changed_modules = []
+    changed_files = []
     included_tags = []
     excluded_tags = []
     if should_only_test_modules:
-        str_test_modules = [m.strip() for m in opts.modules.split(",")]
-        test_modules = [m for m in modules.all_modules if m.name in str_test_modules]
-
         # If we're running the tests in Github Actions, attempt to detect and test
         # only the affected modules.
         if test_env == "github_actions":