1) added new mode LAP: linear algebraic projection, now projection is more than 10 times faster.
2) modified mode tensordecomposition: now it uses LAP also, not any more quadrature.

1) added new mode DGTSR: double Gamow-Teller transition sum rules

1) more pair condensates: now one can order more than one pair condensates in variation

2) tensor decomposition: with a given pair condensate, one can decompose it into different parts with good angular momentum (J M)

3) added modes to skip Pandya transformation, i.e. do things without particle-hole transformation, with brute force

1) optimization:

	src/class.cpp PairCondensateProduct::ovemix11, PairCondensateProduct::ovemix21 are optimized, by introducing intermidiate matrixproducts, the total times of matrix multiplication is reduced. The same is done to cPCP::ovemix11 and cPCP::ovemix21

	src/lapack-gemm.cpp used dgemm in blas to do matrix multiplication. For one multiplication with dim = 40-50, dgemm is about 4-5 times faster than our hand-written matrix-multiplication c++ lines, according to our test runs.

2) expansion on projection:

	now the code can project states our of SEVERAL pair condensates, not any more restricted to one pair condensate.

1) rename: because variation is added, so We rename the code as PVPC, standing for projection + variation + pair condensate
2) code reconstruction: now everything is built within or around a class named "PairCondensateProduct", which stands for < p1^n | xx | p2^n >, xx can be 1- or 2-body operators. In this way it seems easier to track the intermediate matrix products and traces.
3) variation: get a energetically lowest pair condensate by variation process, realized with gsl multifunction fdf_minimizer
4) triaxial projection: coded up the triaxial projection in the traditional quadrature way, with 3 Euler angles.
5) random-M0-pair: generate a proton/neutron random M=0 pair, with given space. This is just a small gadget for convenience.

1) speedup matrix elements to be 7 times faster

1) Rename: because projection is added, so I rename the code as PNPC, with P standing for projection.
2) axial projection: Added angular momentum projection codes, to project states out of pair condensates. So mode "e" is modified as below:
3) mode (e): With given M-scheme pairs, construct condensate configurations with total Mt = 0, project out bases with different J, and then diagonalize, getting eigenvalues and eigenwaves with good J.

1) mode (e): calculate Eg.s. =  < w | H | w > / < w | w >, | w > = P^N >, with given P.
2) mode (u): unit tests on < w | w >, < w | A1^\dagger A2 | w >, < w | Q | w >