feat: add a new ratio config instead of table_data_deserialized_data_bytes

[chienguo]

I hereby agree to the terms of the CLA available at: https://docs.databend.com/dev/policies/cla/

Summary

• Fixes #[Link the issue here]
  feat: add a new ratio config instead of table_data_deserialized_data_bytes #14866

Tests

• Unit Test
• Logic Test
• Benchmark Test
• No Test - Explain why

Type of change

• Bug Fix (non-breaking change which fixes an issue)
• New Feature (non-breaking change which adds functionality)
• Breaking Change (fix or feature that could cause existing functionality not to work as expected)
• Documentation Update
• Refactoring
• Performance Improvement
• Other (please describe):

---

This change is 

[dantengsky]

@chienguo

Thanks for your contribution!

There's a concern that hoping you can consider in this PR:

In some scenarios (such as testing and deploying on cloud services), retaining the table_data_deserialized_data_bytes setting alongside table_data_deserialized_memory_ratio would be more convenient.

Furthermore, if both settings are non-zero, the table_data_deserialized_data_bytes setting should have higher priority.

Similar handling in the settings for spilling may serve as a reference (however, in the scenario of this PR, there's no need to consider settings at the per_core level).

https://github.com/datafuselabs/databend/blob/345cba5aa73ba3e93f4c6372a75fb956812243e4/src/query/service/src/pipelines/builders/builder_sort.rs#L212C1-L239C5

cc @BohuTANG

[BohuTANG]

@chienguo

Thanks for your contribution!

There's a concern that hoping you can consider in this PR:

In some scenarios (such as testing and deploying on cloud services), retaining the table_data_deserialized_data_bytes setting alongside table_data_deserialized_memory_ratio would be more convenient.

Furthermore, if both settings are non-zero, the table_data_deserialized_data_bytes setting should have higher priority.

Similar handling in the settings for spilling may serve as a reference (however, in the scenario of this PR, there's no need to consider settings at the perf_core level).

https://github.com/datafuselabs/databend/blob/345cba5aa73ba3e93f4c6372a75fb956812243e4/src/query/service/src/pipelines/builders/builder_sort.rs#L212C1-L239C5

cc @BohuTANG

@dantengsky Your suggestion looks great to me.
@chienguo Sorry for issue #14866 , it seems it wasn't clear enough, @dantengsky added clarity.

[chienguo]

@dantengsky, already made the original one back, and have dealt with the priority

[dantengsky]

@chienguo LGTM & thanks

[BohuTANG]

@chienguo 👍 . Thanks for your contribution.

[chienguo]

@BohuTANG, @dantengsky, thanks for your help