Implement createToken in Syntax

src/syntax.coffee

@@ -4,6 +4,7 @@ _ = require 'underscore-plus'
 {GrammarRegistry, ScopeSelector} = require 'first-mate'
 
 {$, $$} = require './space-pen-extensions'
+Token = require './token'
 
 ### Public ###
 module.exports =
@@ -26,6 +27,8 @@ class Syntax extends GrammarRegistry
   serialize: ->
     {deserializer: @constructor.name, @grammarOverridesByPath}
 
+  createToken: (value, scopes) -> new Token({value, scopes})
+
   addProperties: (args...) ->
     name = args.shift() if args.length > 2
     [selector, properties] = args

src/tokenized-buffer.coffee

@@ -203,7 +203,6 @@ class TokenizedBuffer extends Model
     lineEnding = @buffer.lineEndingForRow(row)
     tabLength = @getTabLength()
     { tokens, ruleStack } = @grammar.tokenizeLine(line, ruleStack, row is 0)
-    tokens = (new Token(token) for token in tokens)
     new TokenizedLine({tokens, ruleStack, tabLength, lineEnding})
 
   # FIXME: benogle says: These are actually buffer rows as all buffer rows are