subresample: Get a resampled subsample of blm and lexpit estimates.
Description
This method computes a matrix of n.boot subsampled resamples of the regression model. If m is equal to the dataset size, then a standard bootstrap is performed. Otherwise m of n observations, drawn without replacement, are used to re-estimate the model. If m is null, than a stochastic m is selected based on a uniform $\alpha$ and $m=n^\alpha$. The minimum value of $\alpha$ is chosen so that the minimum m is large enough that the expected number of events at the minimum subsample size is 10 per model parameter.