JVM native boundary: has FFM (Project Panama) been considered instead of/alongside JNI?

[Ma77Ball]

The docs describe the engine passing the translated ExecutionPlan to the native side through JNI calls (introduction page, "Apache Auron's Solution" and the "JNI gateway" component). I searched existing issues and discussions and did not find this raised before; the closest is #54, which batches calls to amortize JNI overhead.

Has the Foreign Function & Memory API (FFM / JEP 454, formerly Project Panama) been considered for this boundary, given its lower per-call overhead than JNI? A few specifics:

1. Is JNI a deliberate long-term choice, or a starting point? Since per-crossing cost is already amortized by passing whole Arrow batches rather than per-row, JNI overhead may not currently be a bottleneck. Is that the reasoning?
2. Is the choice driven by supported-JDK range? FFM finalized in JDK 22, while Auron targets JDK 8/11/17, so I assume backward compatibility is a factor.
3. Is FFM on the roadmap for a future implementation (e.g. an optional path gated on a newer JDK), or not currently planned?

Happy to open an issue/AIP if there is interest.