chore: replace Vec<bool> with bitvec (#2360)

bitvec is 8x smaller than Vec<bool>, tests show there's no performance
diff

---------

Signed-off-by: BubbleCal <[email protected]>

Cargo.toml

@@ -77,6 +77,7 @@ half = { "version" = "2.4.1", default-features = false, features = [
     "num-traits",
     "std",
 ] }
+bitvec = "1.0.1"
 bytes = "1.4"
 byteorder = "1.5"
 clap = { version = "4", features = ["derive"] }

rust/lance-index/Cargo.toml

@@ -19,6 +19,7 @@ arrow-schema.workspace = true
 arrow-select.workspace = true
 async-recursion.workspace = true
 async-trait.workspace = true
+bitvec.workspace = true
 datafusion-common.workspace = true
 datafusion-expr.workspace = true
 datafusion-physical-expr.workspace = true

rust/lance-index/src/vector/graph.rs

@@ -9,6 +9,7 @@ use std::collections::BinaryHeap;
 use std::sync::Arc;
 
 use arrow_schema::{DataType, Field};
+use bitvec::vec::BitVec;
 use lance_core::Result;
 
 pub mod builder;
@@ -150,15 +151,15 @@ pub trait Graph {
 
 /// Array-based visited list (faster than HashSet)
 pub struct Visited<'a> {
-    visited: &'a mut [bool],
+    visited: &'a mut BitVec,
     recently_visited: Vec<u32>,
 }
 
 impl<'a> Visited<'a> {
     pub fn insert(&mut self, node_id: u32) {
         let node_id_usize = node_id as usize;
         if !self.visited[node_id_usize] {
-            self.visited[node_id_usize] = true;
+            self.visited.set(node_id_usize, true);
             self.recently_visited.push(node_id);
         }
     }
@@ -172,22 +173,22 @@ impl<'a> Visited<'a> {
 impl<'a> Drop for Visited<'a> {
     fn drop(&mut self) {
         for node_id in self.recently_visited.iter() {
-            self.visited[*node_id as usize] = false;
+            self.visited.set(*node_id as usize, false);
         }
         self.recently_visited.clear();
     }
 }
 
 #[derive(Debug, Clone)]
 pub struct VisitedGenerator {
-    visited: Vec<bool>,
+    visited: BitVec,
     capacity: usize,
 }
 
 impl VisitedGenerator {
     pub fn new(capacity: usize) -> Self {
         Self {
-            visited: vec![false; capacity],
+            visited: BitVec::repeat(false, capacity),
             capacity,
         }
     }