[FEA] Parallelize documentation post-processing

[bdice]

Documentation deployment takes more than one hour to run. It may make sense to parallelize the CPU operations that post-process the documentation files. I would guess that this is the slow step that would benefit from parallelization:

docs/ci/customization/customize_docs_in_folder.sh

Lines 41 to 51 in 1d9a390

	for FILE in $(grep "${JTD_SEARCH_TERM}\|${DOXYGEN_SEARCH_TERM}\|${PYDATA_SEARCH_TERM}" -rl \
	--include=\*.html \
	--exclude-dir=stable \
	--exclude-dir=nightly \
	--exclude-dir=latest \
	--exclude-dir=legacy \
	--exclude-dir=cudf-java \
	${FOLDER_TO_CUSTOMIZE} ); do
	python ${SCRIPT_SRC_FOLDER}/customize_doc.py $(realpath ${FILE})
	echo "" # line break for readability
	done

[bdice]

This may become irrelevant with the ongoing work to unify rapids.ai and docs.rapids.ai.