CAY-2847 Improve duplicate select column detection when using order by

[Jugen]

The current implementation has three problems:

1. It fails when a column has an alias
2. It builds the whole SQL column string before comparing
3. It doesn't seem very robust in preventing false results

This PR replaces the current implementation with a visitor pattern that fails fast.
The visitor extracts each node's parts into a List which is progressively compared to the parts extracted from the order node.
Each item must match in position and content in both lists for them to match and will abort comparing as soon as possible.

[stariy95]

This seems a better solution than what we already have, so I'll merge it as soon I'll be back at Cayenne. There's a minor wish, but that's optional.

Looking at the whole feature, I dream that we'll have a proper comparison API for the SQL nodes one day, and that would make our life so much easier 😄

[Jugen]

I think that there are two tests needed but I don't know how to add them to the current test setup in OrderingStageTest

1. Test that no duplicate column is added for an aliased select column
2. Test that no duplicate column is added for a complex operator column

These are not critical tests, just sanity checking that the code does what is expected.

[Jugen]

I've rebased & squashed commits for this PR, and updated the release-notes.