Gridvoting

Includes tested, improved stationary distribution solver for Markov Chains.

Handles top cycles correctly, reducing to zero the probability of the unreachable alternatives.

Raises `RuntimeError` in cases where the power method does not converge or the algebraic method has issues.

added method grid.extremes

`(min_z,min_z_points,max_z,max_z_points) = grid.extremes(z,valid)`

**this breaks code using grid.as_xy_vectors() **
replace function `grid.as_xy_vectors()` with property `grid.points`

rewrite summarize_in_context to report points [x,y] and not x and y coordinates to make more human-readable

support user-friendly array input to grid.spatial_utilities and grid.within_triangle

support cases with cores in VotingModel.summarize_in_contest and include core tests in the replication testing module

adjust tests

added VotingModel.summarize_in_context() for grid/valid based statistical summary;
added MarkovChainCPUGPU.diagnostic_metrics()

adjust VotingModel.plots usage of embedding function; improve README