[Feature] Optimize splits generator for unaware bucket mode

### Search before asking

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

### Motivation

In stream mode, the unaware bucket mode will use the `FIFOSplitAssigner`, So we could avoid Bin-Packing serveral files into one split. IMO, it has two benefit:

- In the `FIFOSplitAssigner`, we have a work-stealing mechanism that allows for higher total throughput as the split size decreases.
- We can avoid the issue of skewed files by comparing two files with similar row counts but very different file sizes.

### Solution

One file one split for unaware bucket mode

### Anything else?

_No response_

### Are you willing to submit a PR?

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