Lazily load text files
marK currently does not lazily load text files, this causes a bad perfomance when editing big files. Example file to reproduze it: https://github.com/json-iterator/test-data/blob/master/large-file.json (24M)
(it is advisable to try bigger files too)