Implement `summary` function

[gaow]

As mentioned in passing in #36, function summary can now be applied to a susieR fit to get more concise / polished output. The idea is that by default users should only need to run susie() then use coef, predict and summary to access all information they need. Plus perhaps additionally susie_plot(). In other words we might want to mimic how lm behaves.

I have already implemented a version of summary, to be demonstrated in upcoming vignettes. We can add more to it as we see fit.

[pcarbo]

@gaow Your summary function is a good start. A couple suggestions:

• Make sure it has the same argument names as the default summary function, summary(object, ...).

• By convention, the result should be class c("summary.susie","list"), and then you should have a separate print.summary.susie function to turn this output into a nice, printable summary that ideally is short (yet informative!). Ideally, the output of summary.susie should also be fairly concise (hence, "summary"). See summary.varbvs.R in my varbvs package for an illustration of this.

[gaow]

@pcarbo good suggestions particularly on the print pointer. I'll keep this ticket open to revisit.