[spark] Limit the scope and semantics of startup.mode

### Search before asking

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

### Motivation

startup.mode should only be used in spark streaming scenario, never be used in batch read.

### Solution

_No response_

### Anything else?

_No response_

### Willingness to contribute

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

Start by locating the Spark integration code that handles startup.mode, then compare its streaming and batch-read paths. Confirm where the setting is currently accepted and add or adjust coverage for the intended streaming-only semantics. Done means startup.mode applies to Spark streaming and is not used in batch reads.

Written by the indexing model from the issue text.