streamline default factor naming across qmethod

[maxheld83]

Here's a thought:

• unrotated principal components -> PCi
• automatically rotated principal components -> RCi (because those are actually no longer principal components, but just components)
• manually rotated principal components -> mRCi

This would not make a huge difference, but it would be more strictly correct.

We could then (properly) reserve Fi for factors, if and when they are actually available via Centroid #15 or PAF #20 .

[maxheld83]

here's a thought how to go about this systematically / elegantly:

• have qmethod() and all the downstream functions work entirely on indices, without relying on, or producing names
• then always run q.fnames over the results to give names
  • either the automatically created, but appropriate names as in the above
  • or manually created names from q.fnames or q.mrot.do or wherever
• implement names by taking them from the rot.mat and have them trickle down from there.

This might need a bit of refactoring though, not quite clear whether it's worth the effort.

Open question: how to deal with flagging; should that be in the names?

Properly implementing this might also require / should be done in tandem with #277