[FIP-28] Support historical partitions for datalake-enabled tables

### Search before asking

- [x] I searched in the [issues](https://github.com/apache/fluss/issues) and found nothing similar.

### Motivation

FIP-28 proposes improving historical partition support for datalake-enabled tables. Today, once a Fluss auto partition expires, the data may still exist in the lakehouse storage, but Fluss does not provide a complete user-facing path to keep operating on that historical partition.

This umbrella issue tracks the work needed to make historical partitions usable after they are no longer active Fluss partitions. The goal is to cover both read-side point lookup and write-side support, while keeping the implementation split into reviewable tasks.

### Solution

Track the implementation in two focused issues:

- [x] #3632
- [x] #3633

The point lookup work should allow lookup requests for expired partitions to resolve the target historical lake partition and read matching rows from the lake table. The write-side work should define and implement the behavior for writing records that target historical partitions.

### Anything else?

Related design: FIP-28 historical partition support for datalake-enabled Fluss tables.

### Willingness to contribute

- [x] I'm willing to submit a PR!