Enhanced Features in 'Import Into' SQL Statement

[D3Hunter]

Feature Request

Is your feature request related to a problem? Please describe:

Describe the feature you'd like:

The 'import into' SQL was introduced in version 7.2, integrating the Lightning local backend (physical import mode). In version 7.2, it was released as an experimental feature and Only support 1 TiB source data and can’t resume from breakpoint. So we have to deliver the following new features to make "import into " more powerful.

This newly added feature in version 7.4 ：

1. Support "S3 global sort" to avoid problems caused by data range overlap, to make import more stable.
2. support parameters like "SPLIT_FILE", and support import compressed source files like .gzip, .gz, .zstd, .zst and .snappy.
3. Expected to support data imports of up to 100 TiB.（The final size depends on the results of testing.）
4. The import performance can reach up to 1 TiB per hour.

Describe alternatives you've considered:

Teachability, Documentation, Adoption, Migration Strategy:

[D3Hunter]

this part is release in v7.4.0, so close it