Investigate translation with headless browsers for better js-based loading website support

[adomasven]

Translation solely via XHR is an issue on some translators (currently 70+ translators have test cases with defer: true), primarily because additional loading via JS happens on those websites (e.g. https://www.fao.org/documents/card/en?details=cc0461en)

• We would need to mark translators that require running via a headless browser. Some only need to wait for JS on search pages, which is not important for TS.
• A headless browser would be spun up via puppeteer for those translators and wait some set time before running translation.
• An investigation into resource usage is required
• An investigation into sandboxing cookies/localstorage is required between translation runs
• An investigation into running on AWS lambda is required

[adomasven]

Related zotero/zotero#4255