proposed improvements in Bayes reports

[jbrezmorf]

Chain plot:

• add parameter names
• merge phasis into a single plot just separated by a vertical line, better to compare
• Can we have some numerical quantity measuring the convergence of the chains?

Parameter pair histograms:

• use the lower half to provide Gaussian correlation in form of an ellipse and the value of the correlation coefficient
  can also contain more then single best fit, possibly 5 or 10 best fits with likelihood value as size or intensity of the point.
• unify ranges of blue and gray plots, seems given by posterior STD
• able to prescribe ranges explicitly, for comparison with other inversions
• larger parameter description, move to left in order to be aligned with columns
• lager and more detailed ranges only on the left side, omit values on axis inside of the table, only use major and minor ticks