[Feature] Presto/Trino has poor read performance and needs to be enhanced

### Search before asking

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

### Motivation

After the launch of the Presto333 support for reading Paimon tables, many users in our company reported that querying Paimon tables through Presto333 was too slow. After repeated testing and check in multiple SQL scenarios, it was found that this issue does exist (note: compared to querying Hive tables, Presto queries Paimon tables take 3-6 times longer, but Spark reading performance is much better). The following is a screenshot of the check result:

We hope to enhance the performance of Presto333 in querying Paimon tables (similar to Spark3): querying Paimon tables takes almost as much time as directly querying Hive tables

### Solution

_No response_

### Anything else?

_No response_

### Are you willing to submit a PR?

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

Start by reproducing the reported Presto333 queries against Paimon tables and compare their timings with equivalent Hive and Spark queries. The issue names no files or tests; done means identifying and implementing a focused performance improvement that brings Presto333 query time close to Hive query time.

Written by the indexing model from the issue text.