fix: Emit equality conditions for Substrait CASE base expressions

[namanjain24-sudo]

Which issue does this PR close?

• Closes Substrait producer emits non-boolean IfThen clauses for CASE with a base expression #25190.

Rationale for this change

Substrait's IfThen has no base expression. The spec writes every clause as if <boolean expression> then <result expression>.

The producer used IfThen for CASE <base> WHEN <value> anyway, through a convention private to DataFusion: a first clause holding the base in if with then left unset, then one clause per WHEN carrying the raw WHEN operand. SELECT CASE a WHEN 1 THEN 'x' WHEN 2 THEN 'y' ELSE 'z' END came out as three clauses whose conditions are the field reference a and the literals 1 and 2. None is boolean, and the first has no result. Our consumer reads the convention back, so round trips were unaffected and no test failed. Another engine sees clauses it cannot evaluate.

What changes are included in this PR?

from_case now emits one clause per WHEN, with <base> = <value> as the condition. That is the desugaring from_between already uses for BETWEEN, and it is how the spec describes a switch: an if expression whose conditions are all equality against the same value. The searched form is unchanged. Meaning is preserved, because DataFusion matches a base CASE with the same equality kernels = lowers to.

Left out deliberately:

• SwitchExpression itself. Its IfValue.if is a Literal, so it cannot hold CASE a WHEN b + 1, and our consumer answers not_impl_err!("Switch expression not supported").
• The consumer, which still accepts the old encoding, so plans written by older DataFusion versions keep reading.

The cost is that the base expression is repeated once per WHEN arm, and evaluated per arm.

What is the testing strategy for this PR?

case_with_base_expression_emits_equality_conditions inspects the produced protobuf instead of round-tripping, because the consumer understands the old encoding and a round trip cannot catch this. case_with_base_expression moves to assert_expected_plan to record the new shape, still asserting the schema is unchanged.

Are there any user-facing changes?

A base CASE now serialises to equal calls with a boolean output type. No Rust API changes. Inside DataFusion the plan round trips into the equivalent searched CASE, with the same schema and results. A consumer that implemented the old convention sees the new form, which is valid Substrait.

[namanjain24-sudo]

@gabotechs @kosiew when you have a moment, would one of you be able to trigger the workflows on this PR? The contributor guide notes a committer has to do that for a new contributor, and no checks have run here yet.

Locally this passes cargo test -p datafusion-substrait (272 tests), cargo xtask ci step test substrait, and the rust_clippy.sh, rust_fmt.sh, typos_check.sh and rust_docs.sh steps of the lint suite.

On the change itself: the one judgement call is desugaring CASE <base> WHEN <value> into <base> = <value> conditions rather than emitting SwitchExpression. I went that way because SwitchExpression.IfValue.if is a Literal, so it cannot express a non-literal WHEN operand, and our consumer currently answers not_impl_err!("Switch expression not supported"), so emitting it would break our own round trip. The trade-off is that the base expression is now repeated once per WHEN arm. I described both in #25190 and would be glad to take the PR in a different direction if you'd prefer.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 75.00000% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 81.93%. Comparing base (f8cc678) to head (b87b857).

Files with missing lines	Patch %	Lines
...ubstrait/src/logical_plan/producer/expr/if_then.rs	75.00%	0 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #25191      +/-   ##
==========================================
- Coverage   81.93%   81.93%   -0.01%     
==========================================
  Files        1133     1133              
  Lines      423529   423529              
  Branches   423529   423529              
==========================================
- Hits       347028   347018      -10     
- Misses      55910    55920      +10     
  Partials    20591    20591              

☔ View full report in Codecov by Harness.
📢 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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[gabotechs]

Hi @namanjain24-sudo, thanks for the PR!

Could I ask to summarize and condense a bit the PR description? Feel free to keep anything that is not already implicit in the code itself, aim for a 2 or 3 min read summary. That would help reviewers better digest the PR. Thanks!

[namanjain24-sudo]

Thanks, trimmed it. What I cut (the local test log, and the longer reasoning for leaving SwitchExpression alone) is written up in #25190 if it comes up in review.

On the red check, I don't think it's this PR: it's cargo test datafusion-cli, which has no dependency on datafusion-substrait. MinIO withdrew minio/minio from Docker Hub on the 11th, so the storage integration tests panic on startup, and my branch is based on a commit from before #25216 moved that pull to quay.io. Rebasing onto main should clear it, happy to do that whenever, though it will probably need another workflow approval from you.