Token Variable in Rule Based Matching

[delzac]

I'm thinking about contributing Token Variable to Rule Based Matcher.

Currently, there's no good way to match for reduplications in a general way. For example:
He was very very energetic and going at really really fast speed.

spaCy's rule based matcher has difficulty matching for reduplication because it is currently not possible to express tokens with reference to another matched token.

# adverb, same as first match, adjective (matches either very very energetic or really really fast)
pattern = [{'POS': 'ADV'}, {__SAME_AS_FIRST_TOKEN__}, {'POS': 'ADJ'}] 

While reduplication is uncommon in english, in many other languages like Bahasa Melayu, it is how plurals are expressed.

Also, with Token Variables will allow for greater expressiveness in matching of patterns such as coast to coast or end to end

So, i would like to find out if the maintainers would be keen to accept a PR on this?