Idris in ide-mode sends wrong sexp expression (make case clause)

[justjoheinz]

Steps to Reproduce

Given the following program:

module Main

myFun : ?hole -> String
myFun x = ?myFun_rhs

• move your cursor to ?myFun_rhs and get the type of the hole (Atom: Ctrl-Alt-t)

Expected Behavior

idris sends correct s-exp expression with no duplicates.

Observed Behavior

idris responds with the s-exp expression:

(
(2 1 ((:name "x") (:decor :bound) (:implicit :False)))
-> (7 4 ((:name "Main.hole") (:implicit :False) (:key "a==") (:decor :metavar) (:type "Type") (:namespace "Main")))
-> (6 5 ((:name "Main.hole") (:implicit :False) (:key "a==") (:decor :metavar) (:type "Type") (:namespace "Main")))
(51 9 ((:name "myFun_rhs") (:implicit :False) (:key "AAAAAAAAAAAJbXlGdW5fcmhz") (:decor :metavar) (:type "?hole -> String")))
(63 6 ((:decor :type) (:type "Type") (:doc-overview "Strings in some unspecified encoding") (:name "String"))))) 3)

which contains two overlapping definitions of the metavar hole (marked by ->)
this leads to problems in IDEs such as Atom, which try to implement a reasonable UI for this.

See idris-hackers/atom-language-idris#195 for screenshots