Can't merge non-disjoint spans. '🇸' is already part of tokens to merge.

First off, thanks for this amazingly useful library! Really takes the sting out of working with social media texts.

I found something that looks like a bug occurring with multi-codepoint emoji’s. Repro case:

import en_core_web_sm
from spacymoji import Emoji

nlp = en_core_web_sm.load()
em = Emoji(nlp)

em(nlp(u"🇺🇸🇦🇷"))

This raises the exception shown in the title. Initializing Emoji with merge_spans=False circumvents this.

Let me know if I can help!

[scribu]

#8 is my attempt at fixing this. Please try it out and let me know if you encounter other edge cases.

[i-like-the-idea]

Can someone merge the fix please?