Add posterior mean or mode or MAP curve to plot_curve_hdi

[drbenvincent]

The Transformation class has some plot methods such as:

• plot_curve_hdi just plots the HDI
• plot_curve plots HDI and some random samples

But it would be nice if we could allow the user to also plot a singular curve that represents out 'best guess'.

If the curve function is univariate then this is easy because we could just plot the curve for the posterior mean or median. If the curve function has multiple parameters then because the posterior could be a more complex shape than a simple bivariate normal (it could be curved), then for the curve to have meaning then we'd need to find the posterior mode.

[wd60622]

If functions in pymc_marketing.mmm.plot (soon to be pymc_marketing.plot) get this change, then they will be spread everywhere they are used. Maybe an option in the plot_curve function