chore: add a setting for inlist to join (#15024)

Author: xudong963

src/query/settings/src/settings_default.rs

@@ -290,6 +290,12 @@ impl DefaultSettings {
                     mode: SettingMode::Both,
                     range: Some(SettingRange::Numeric(0..=u64::MAX)),
                 }),
+                ("inlist_to_join_threshold", DefaultSettingValue{
+                    value: UserSettingValue::UInt64(1024),
+                    desc: "Set the threshold for converting IN list to JOIN.",
+                    mode: SettingMode::Both,
+                    range: Some(SettingRange::Numeric(0..=u64::MAX)),
+                }),
                 ("enable_bloom_runtime_filter", DefaultSettingValue {
                     value: UserSettingValue::UInt64(1),
                     desc: "Enables runtime filter optimization for JOIN.",

src/query/settings/src/settings_getter_setter.rs

@@ -285,6 +285,10 @@ impl Settings {
         Ok(self.try_get_u64("join_spilling_partition_bits")? as usize)
     }
 
+    pub fn get_inlist_to_join_threshold(&self) -> Result<usize> {
+        Ok(self.try_get_u64("inlist_to_join_threshold")? as usize)
+    }
+
     pub fn get_bloom_runtime_filter(&self) -> Result<bool> {
         Ok(self.try_get_u64("enable_bloom_runtime_filter")? != 0)
     }

src/query/sql/src/planner/semantic/type_check.rs

@@ -380,7 +380,9 @@ impl<'a> TypeChecker<'a> {
                 not,
                 ..
             } => {
-                if self.ctx.get_cluster().is_empty() && list.len() >= 1024 {
+                if self.ctx.get_cluster().is_empty()
+                    && list.len() >= self.ctx.get_settings().get_inlist_to_join_threshold()?
+                {
                     if *not {
                         return self
                             .resolve_unary_op(*span, &UnaryOperator::Not, &Expr::InList {