[ bug:racket ] Repro the issue with epsilon.0

For some reason, having `epsilon.0` from `(require math/flonum)`
anywhere in the Racket support library trips the totality checker,
regardless of whether the code is used or not. Operations from
`math/flonum` (e.g. `fl/`) do not reproduce this behaviour.

See idris-lang#3116 for the full context.

src/Compiler/Scheme/Racket.idr

@@ -56,6 +56,7 @@ schHeader prof libs = fromString """
   (require ffi/unsafe ffi/unsafe/define) ; for calling C
   \{ ifThenElse prof "(require profile)" "" }
   (require racket/flonum)                ; for float-typed transcendental functions
+  (require math/flonum)                  ; for flonum constants
 
   """ ++ libs ++ """
 

support/racket/support.rkt

@@ -37,6 +37,20 @@
       (if (> b 0) (+ r b) (- r b))
       r)))
 
+
+; flonum constants
+
+;; /!\ This code trips the totality checker, even when it is never referenced in
+;; /!\ Idris2, but only on the file `libs/papers/Data/Description/Regular.idr`??
+
+(define (blodwen-flonumEpsilon)
+  epsilon.0)
+
+;; /!\ End of cursed code
+
+
+; Bits
+
 (define bu+ (lambda (x y bits) (blodwen-toUnsignedInt (+ x y) bits)))
 (define bu- (lambda (x y bits) (blodwen-toUnsignedInt (- x y) bits)))
 (define bu* (lambda (x y bits) (blodwen-toUnsignedInt (* x y) bits)))