Revive AugmentedDirac

[blegat]

#49 made the tests fail so I commented them out because I don't understood why they would work. Because AugmentedDirac does not have the same element type than DiracBasis, how could coeffs work ?
See

StarAlgebras.jl/test/perm_grp_algebra.jl

Lines 34 to 59 in 7760127

	# @testset "Augmented basis" begin
	# ad = SA.AugmentedBasis(db)
	# @test SA.mstructure(ad) == SA.AugmentedMStructure(SA.mstructure(db))
	# @test ad[SA.Augmented(h)] isa SA.Augmented
	# @test sprint(show, ad[SA.Augmented(h)]) == "(-1·()+1·(1,2,4,5))"
	#
	# @test !(h in ad)
	# @test SA.Augmented(h) in ad
	#
	# IG = SA.StarAlgebra(G, ad)
	#
	# axcfs = SA.coeffs(x, basis(IG))
	# aycfs = SA.coeffs(y, basis(IG))
	# azcfs = SA.coeffs(z, basis(IG))
	# ax = SA.AlgebraElement(axcfs, IG)
	# ay = SA.AlgebraElement(aycfs, IG)
	# az = SA.AlgebraElement(azcfs, IG)
	#
	# @test coeffs(ax * ay) == SA.coeffs(x * y, basis(IG))
	# @test coeffs(ax * az) == SA.coeffs(x * z, basis(IG))
	# @test SA.aug(ax) == 0
	# @test star(ax) * star(ay) == star(ay) * star(ax)
	#
	# @test length(ad) == length(db) - 1
	# @test Set(ad) == Set(SA.Augmented(g) for g in db if !isone(g))
	# end

[kalmarek]

The idea is that the user provides a way of writing one basis elements as coeffs in the other basis, so that we can translate between those bases.

Here the translation is

1 → 1
x → 1 - x

But I'd need to walk through the whole code of this and see what can be salvaged