Investigate enabling new model as the default/only model for language detection

Testing #143576 Right now, @JacksonKearl's model is the backup to the Guesslang model used in language detection. We should enable some way to make it be the only model used (and potentially replace Guesslang). Why? This feature was initially intended to help new users who are pasting in code snippets - to set them up for success by guessing the language and setting it so that they can be taken down the happy path of installing the right extension. I think as long as we support the most popular languages (that @JacksonKearl's model already does I believe) that will have the biggest impact. I'd argue that Jackson's model that supports less languages but is able to guess off smaller inputs **is more useful** than what we already have which supports more languages but requires more text to guess correctly.