Update the mixed state decomposition method

[garymooney]

Update the method of decomposing mixed states to pure states. At the moment we just do an eigen decomposition. To give the main pure state more meaning, it should be what the state would be if it was simulated without noise. Then the corresponding pure states representing the noise would be calculated as being orthonogoal to the each other and the main pure states.

[garymooney]

Changing the decomposition to include noiseless simulation as the main coherent state might cause issues with noise models that have coherent noise, since the coherent noise will not be included and displayed as part of the main pure state. While, the current decomposition strategy of using an eigendecomposition still gives a good indication of the coherent state.

With this realisation, this issue will be put aside (set to low priority) until a better alternative is presented that handles coherent error well.

A potential approach could be to include only coherent errors in the "noiseless" simulation when obtaining the coherent state.