Warn when applying ordering to a query with pre-existing order

[tomekowal]

Hey!

This PR attempts to mitigate issue #522
I was thinking about removing the pre-existing order_by from queries, but Ecto docs mention that reading the internal structure of the query is somewhat OK, but modifying is a no go.

https://hexdocs.pm/ecto/Ecto.Query.html#__struct__/0

Users of Ecto must consider this struct as opaque and not access its field directly. Authors of adapters may read its contents, but never modify them.

Flop is not exactly an adapter, but I think similar logic applies.

Using the internal API of Ecto seems to me as a good trade for avoiding the confusion of accidentally passing a query with pre-existing order_by which leads to a very weird behaviour and a lot of confusion.

We could potentially be more strict and even raise when we have a query with pre-existing order_by, but that would be a breaking change.

Let me know what you think, I am happy to make any corrections you find necessary.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 87.59%. Comparing base (c2aed56) to head (822fcdd).
Report is 89 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #524      +/-   ##
==========================================
- Coverage   87.67%   87.59%   -0.09%     
==========================================
  Files          15       15              
  Lines         933      935       +2     
==========================================
+ Hits          818      819       +1     
- Misses        115      116       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[woylie]

Thank you!