provide support for multilanguage packages and projects

[ThierryO]

This is PR adds a lot a functionality to spelling. Basically because I needed a easy way to spell R markdown projects. We often use multiple languages in a single project. E.g. a report written in Dutch with an English abstract and README. Or a package with function and documentation in English and a report template in Dutch.

I'm happy to do some more work on this PR if you want some more streamlining.

Changes

• existing functions work in the same way as they did before.
• the summary output gained a lang variable and displays the language used for spell checking.
• added spell_check_project() to check a project with (R)markdown files
• spell checking a project or a multi language package writes a spell_check.json to store the language information. Single language packages relay on the information from DESCRIPTION.
• change_language() allows to set a non-default language for files or folders (and their childern). Update DESCRIPTION to change the default language of a package. Use spell_check_project() with the lang argument to set or change the default language of markdown project.
• Using change_language(lang = "ignore") will ignore a file or folder when spell checking.
• parse_text_md() tries to ignore math and yaml field names. It also ignore lines containing <!-- spell-check: ignore --> or lines between <!-- spell-check: ignore:start --> and <!-- spell-check: ignore:end -->. This was required to ignore some more elaborate math examples which commonmark::markdown_xml() split over multiple lines.
• spell_check_package() gains an inst = FALSE argument. Setting inst = TRUE will include markdown and sweave files in the inst folder. I'd prefer to use inst = TRUE as default, but that might change the spell check results for some packages.
• The default language uses inst/WORDLIST both for packages and projects. Alternative languages use the normalised language as suffix. E.g. inst/WORDLIST_nl_BE.

Tested with

• https://github.com/inbo/multimput (single language package)
• https://github.com/inbo/abvanalysis/tree/meetnetten (multi language package with rmd in inst)
• https://github.com/inbo/patrijs-steekproefkader/tree/rapport (multi language project)

[jeroen]

Thank you for the contribution. This is a very big PR, I am not sure where to start reviewing this.

Could perhaps (part of) this work as a separate package which depends on spelling? Or does it really need to change the internals of spelling?

[ThierryO]

Maybe it is easier for you if I split this over multiple PR's.

1. check markdown and sweave file in inst folder (by default or optional?)
2. ignore math and ignore patterns
3. multi-language support for packages
4. spell checking for rmd projects with multi-language support

This functionality makes spelling IMHO even more useful. I don't think the users would benefit from a 'competing' spell checking package.