check thin argument in getPredictiveIntervals about bayesiantools

[TankredO]

The thinning in this function works differently than the thinning in getSample etc.. Maybe it would make sense to abstract the thinning into an utility function and use it everywhere.

[florianhartig]

I don't think that's necessary, I think thin was simply the wrong word for this, in line with the logic of getSample this could simply be renamed into nSamples.

While you're at it, however, could you check

a) that this works

b) what happens if nSamples is smaller than number of samples available in the chain (might be necessary to catch this in get Sample)

[TankredO]

Should be save now: 3b4db45.
Also, thin was renamed to numSamples to be consistent with getSample.