Fix GPy.priors.InverseGamma (#903)

* fixed InverseGamma prior: beforehand, it was a child class of Gamma but it defined a broken __new__ method of its own. Now, it just inherits Gamma's __new__; also added a test that ensures the InverseGamma can be instantiated and integrated into a GPy model

* overwrote misleading inherited methods in InverseGamma, deleted unnecessary repeated code

GPy/core/parameterization/priors.py

@@ -371,24 +371,17 @@ class InverseGamma(Gamma):
     """
     domain = _POSITIVE
     _instances = []
-    def __new__(cls, a=1, b=.5): # Singleton:
-        if cls._instances:
-            cls._instances[:] = [instance for instance in cls._instances if instance()]
-            for instance in cls._instances:
-                if instance().a == a and instance().b == b:
-                    return instance()
-        o = super(Prior, cls).__new__(cls, a, b)
-        cls._instances.append(weakref.ref(o))
-        return cls._instances[-1]()
-
-    def __init__(self, a, b):
-        self._a = float(a)
-        self._b = float(b)
-        self.constant = -gammaln(self.a) + a * np.log(b)
 
     def __str__(self):
         return "iGa({:.2g}, {:.2g})".format(self.a, self.b)
 
+    def summary(self):
+        return {}
+
+    @staticmethod
+    def from_EV(E, V):
+        raise NotImplementedError
+
     def lnpdf(self, x):
         return self.constant - (self.a + 1) * np.log(x) - self.b / x
 
@@ -398,7 +391,6 @@ class InverseGamma(Gamma):
     def rvs(self, n):
         return 1. / np.random.gamma(scale=1. / self.b, shape=self.a, size=n)
 
-
 class DGPLVM_KFDA(Prior):
     """
     Implementation of the Discriminative Gaussian Process Latent Variable function using