[SPARK-51820][SQL][CONNECT] Address remaining issues for new group/order by ordinal approach

[mihailotim-db]

What changes were proposed in this pull request?

This is a followup to #50606, to address remaining issues in fixed-point analyzer and parser. This PR does the following:

1. Set correct origin in SparkConnect and Dataframe API
2. Handle remaining cases for order by ordinal in Dataframe API
3. Add a test suite for Dataframe use cases since we are missing those

Why are the changes needed?

To address current issue with UnresolvedOrdinal

Does this PR introduce any user-facing change?

No

How was this patch tested?

Added a new test suite and existing tests.

Was this patch authored or co-authored using generative AI tooling?

No

[mihailoale-db]

IIUC we changed the API/approach for functions that explicitly add Sort/Aggregate, but there are other functions/rules that do that implicitly (e.g randomSplit)? @cloud-fan are there any other places where we add those operators that should be handled?

[cloud-fan]

We should check all DataFrame APIs, but randomSplit is not a target as the order by expression is not specified by users.

[mihailoale-db]

@cloud-fan Agree that randomSplit is not an issue but for example you could have a FunctionTableSubqueryArgumentExpression which produces a Sort. My concern is that we have a few other operators/expressions/rules that introduce Sort in the middle of the analysis which are not going to be handled properly (same as ones which we introduce by the API calls).

[mihailotim-db]

@mihailoale-db We are only interested in what happens in early batches (before SubstituteUnresolvedOrdinals was ran) and from what I can see, there are no Sorts anywhere there

[cloud-fan]

thanks, merging to master!