Incorrect aliasing

Here is an exchange between Eric Hodel and me in December 2010:

Me:

Hi Eric,

The bug I started fixing is that when we find
  alias old_met met
  def met ...
then old_met is not an alias for the met method defined there:
it is an alias for a 'met' method defined in an ancestor, so at worst an external alias.
But that means you can only resolve aliases when parsing is complete.
So I did start fixing the bug, but I think it drives me too far,
and as I do not have much time lately,
I think we can keep that for a next version.
Is that OK with you?
-- Thierry

Eric :

Wow, that will be a tough bug to fix.
I admire your ambition and it is fine that this bug waits for a future release.
Since there have been API changes this next release will be RDoc 3.
(...)

At RDoc 8, I think it's time to fix the bug! ^_^

Start by reproducing the alias old_met met followed by def met example described in the issue, then trace RDoc's alias handling and parsing flow. The fix is complete when the alias resolves to the method defined in the current class rather than an ancestor, with regression coverage for this case.