Introduce enable_optimize_read_in_order in CHYT

This PR introduces a new mode in CHYT that aims to mimic the behaviour of [optimize_read_in_order](https://clickhouse.com/docs/en/sql-reference/statements/select/order-by#optimization-of-data-reading) in native ClickHouse. This PR implements #451, albeit using a different approach.
It works by using a sorted pool to produce jobs of exponentially increasing/decreasing sizes, which turn into subqueries that are processed combined sequentially without additional sorting steps. This should allow queries described in #451 to execute without reading the whole table.
Notable modifications with generic impact:
* Support for exponentially increasing job sizes in new sorted pool and related classes.
* Chunk slice fetcher is now used when creating a sorted pool in CHYT.
* New column in query log table exporter.

---
1f9ea932d21d4b328b4be976eb39cdfbc631dffd

Pull Request resolved: ytsaurus/ytsaurus#757

Author: achulkov2

yt/yt/client/table_client/row_base.h

@@ -117,6 +117,17 @@ inline bool IsIntegralType(ESimpleLogicalValueType type)
     }
 }
 
+inline bool IsFloatingPointType(ESimpleLogicalValueType type)
+{
+    switch (type) {
+        case ESimpleLogicalValueType::Double:
+        case ESimpleLogicalValueType::Float:
+            return true;
+        default:
+            return false;
+    }
+}
+
 inline bool IsStringLikeType(ESimpleLogicalValueType type)
 {
     switch (type) {