Add significance testing to MultiplePSD for ensembles.

[khider]

Is your feature request related to a problem? Please describe.
Right now, can't compute a significance test.

Describe the solution you'd like
Some kind of way to assess the significance of the peaks

Additional context
The challenge here is to find one significance to rule them all (i.e., should the autocorrelation coefficient used to generate the surrogates be the average or median of the coefficients for each series; should we compute AR1 surrogates for each series and somehow integrate all of them in the quantiles?)

[khider]

Option - testing style

1. Use the mode/median/max of the lag-1 autocorrelation
2. Run fixed number (default=100) for each lag-1 autocorrelation and use this as the ensemble for the quantile computation.