>> "symmetric deletion, which I’ve ... never seen it used in a production setting"

SeekStorm (search as a service https://seekstorm.com/ ) is using SymSpell in production, and Faroo (peer-to-per-web search) did use it too.

Regarding the limitation, SymSpell on the GitHub repository is more a proof-of-concept of the speed of the algorithm than a production-ready, turn-key product. Think more of an efficient algorithm to pre-filter/reduce the number of candidates. After that, it can be amended with a ranking/prioritization algorithm of your choice.

To improve the precision one could add a Weighted Edit distance giving a higher priority to pairs that are close to each other on the keyboard layout or that sound similar (e.g. Soundex or other phonetic algorithms which identify different spellings of the same sound).

Also, taking the context in a multi-term query or between consecutive queries into account with word embeddings and word vectors is desirable, as described e.g. here: https://blog.usejournal.com/a-simple-spell-checker-built-from-word-vectors-9f28452b6f26