use fmt's %q to encode []byte literals

I now implemented and tested hex encoded strings, binary literals, and fmt.Sprintf("%q", data).
In the case of garble all produce the exact same binary size.

I decided to use %q because it is the simplest, leads to the smallest garbled code file size size
of the three, and is faster at compile time than binary literals.

It looks like:

    var ztN0xdMLL = garbleDecrypt([]byte("\xaf\xbd\x01\\&\x14\xab\xeb\x94\x10Q\xf2H#\xde\x17\a\x8f\x89MmZs\u0088\xcfw\xba?\x9e\xe1\x81\x1eպD\xe1@\xf2\x8d\xe3Ije\xca\bB\xbey\x8b"))

From the fmt docs:

    String and slice of bytes (treated equivalently with these verbs):
    [...]
    %q    a double-quoted string safely escaped with Go syntax

Fixes #40.

strings.go

@@ -6,7 +6,6 @@ import (
 	"go/token"
 	"go/types"
 	"strconv"
-	"strings"
 
 	"golang.org/x/tools/go/ast/astutil"
 )
@@ -285,22 +284,14 @@ func ciphertextStmt(ciphertext []byte) *ast.CallExpr {
 
 // dataToByteSlice turns a byte slice like []byte{1, 2, 3} into an AST
 // expression
-func dataToByteSlice(data []byte) *ast.CompositeLit {
-	var builder strings.Builder
-
-	for _, b := range data {
-		builder.WriteString(strconv.FormatInt(int64(b), 10) + ",")
-	}
-
-	return &ast.CompositeLit{
-		Type: &ast.ArrayType{
-			Elt: &ast.Ident{
-				Name: "byte",
-			},
+func dataToByteSlice(data []byte) *ast.CallExpr {
+	return &ast.CallExpr{
+		Fun: &ast.ArrayType{
+			Elt: &ast.Ident{Name: "byte"},
 		},
-		Elts: []ast.Expr{&ast.BasicLit{
+		Args: []ast.Expr{&ast.BasicLit{
 			Kind:  token.STRING,
-			Value: builder.String(),
+			Value: fmt.Sprintf("%q", data),
 		}},
 	}
 }