+ The skew-symmetric EJA of order `2n` described in Faraut and
+ Koranyi's Exercise III.1.b. It has dimension `2n^2 - n`.
+
+ It is (not obviously) isomorphic to the QuaternionHermitianEJA of
+ order `n`, as can be inferred by comparing rank/dimension or
+ explicitly from their "characteristic polynomial of" functions,
+ which just so happen to align nicely.
+
+ SETUP::
+
+ sage: from mjo.eja.eja_algebra import (ComplexSkewSymmetricEJA,
+ ....: QuaternionHermitianEJA)
+ sage: from mjo.eja.eja_operator import FiniteDimensionalEJAOperator
+
+ EXAMPLES:
+
+ This EJA is isomorphic to the quaternions::
+
+ sage: J = ComplexSkewSymmetricEJA(2, field=QQ, orthonormalize=False)
+ sage: K = QuaternionHermitianEJA(2, field=QQ, orthonormalize=False)
+ sage: jordan_isom_matrix = matrix.diagonal(QQ,[-1,1,1,1,1,-1])
+ sage: phi = FiniteDimensionalEJAOperator(J,K,jordan_isom_matrix)
+ sage: all( phi(x*y) == phi(x)*phi(y)
+ ....: for x in J.gens()
+ ....: for y in J.gens() )
+ True
+ sage: x,y = J.random_elements(2)
+ sage: phi(x*y) == phi(x)*phi(y)
+ True
+
+ TESTS:
+
+ Random elements should satisfy the same conditions that the basis
+ elements do::
+
+ sage: K = ComplexSkewSymmetricEJA.random_instance(field=QQ,
+ ....: orthonormalize=False)
+ sage: x,y = K.random_elements(2)
+ sage: z = x*y
+ sage: x = x.to_matrix()
+ sage: y = y.to_matrix()
+ sage: z = z.to_matrix()
+ sage: all( e.is_skew_symmetric() for e in (x,y,z) )
+ True
+ sage: J = -K.one().to_matrix()
+ sage: all( e*J == J*e.conjugate() for e in (x,y,z) )
+ True
+
+ The power law in Faraut & Koranyi's II.7.a is satisfied.
+ We're in a subalgebra of theirs, but powers are still
+ defined the same::
+
+ sage: K = ComplexSkewSymmetricEJA.random_instance(field=QQ,
+ ....: orthonormalize=False)
+ sage: x = K.random_element()
+ sage: k = ZZ.random_element(5)
+ sage: actual = x^k
+ sage: J = -K.one().to_matrix()
+ sage: expected = K(-J*(J*x.to_matrix())^k)
+ sage: actual == expected
+ True
+