Tokenizer somehow splits the word "id" into 2 sepperate tokens

[CraiyDev]

If you simply call nlp("search for the id 10") you will notice that the word "id" gets split into 2 sepperate tokens named i and d
This behavior happens for en_core_web_md as well as en_core_web_sm, I havent tried more but it seems to be a general problem with tokenization

[adrianeboyd]

By default the English tokenizer settings treat "id" as a variant of "I'd". Here's a comment about how to modify this: #10455 (comment)