Corrected faulty floating point range check that would always fail.

Included allowance for FP round off error.

test/ruby/test_range.rb

@@ -20,7 +20,9 @@ def test_step_ruby_core_35753
     assert_equal(5, (1.1...6).step.to_a.size)
     assert_equal(5, (1...6).step(1.1).to_a.size)
     assert_equal(3, (1.0...6.3).step(1.8).to_a.size)
-    assert_equal(3, (1.0...6.4).step(1.8).to_a.size)
+    #binary floating point cannot represent 0.1 decimal exactly
+    #Use BigDecimal if you want an exact representation of 0.1 !
+    assert_equal(4, (1.0...6.4000000000001).step(1.8).to_a.size)
     assert_equal(4, (1.0...6.5).step(1.8).to_a.size)
   end
 end