[feature] partition/file pruning by indexing

[BohuTANG]

Summary

Impl the apply_index against the expression.

[leiysky]

I have a plan on expressions normalization in #1539 , in purpose of providing ability to do complicated partition pruning(e.g. rewrite a + 1 < 2 to a < 1 so we can directly apply the comparison on partition key).

By the way, it's a very big challeng to do partition prune in a data warehouse, since there may be a large amout of partitions in a single table(says millions), which can bring serious burden to optimizer. The mainstream data warehouse(e.g. Snowflake) has a mechanism like intermidiate partition to reduce the metadata should be read during query optimization.

We can have a more robust design on this in the future.

[BohuTANG]

@leiysky
That's interesting.
Datafuse supports 3-level partition pruning to fast query(Which ClickHouse did, simply and efficiently), there is an introduce in:
https://github.com/datafuselabs/datafuse/blob/master/common/indexing/README.md

This is the pruning after the optimizer,apply index patch will be launched next week.

[zhyass]

/assignme

[zhyass]

Tasks to be done:

• Add RangeFilter for index pruning #2344
• enable VerifiableExpression in fuse table #2362
• Index: RangeFilter support like/not like #2472
• Add monotonic check in FunctionFactory #2343