chore: replace dashmap with clashmap (#10582)

## Problem

Because dashmap 6 switched to hashbrown RawTable API, it required us to
use unsafe code in the upgrade:
https://github.com/neondatabase/neon/pull/8107

## Summary of changes

Switch to clashmap, a fork maintained by me which removes much of the
unsafe and ultimately switches to HashTable instead of RawTable to
remove much of the unsafe requirement on us.

proxy/src/cache/endpoints.rs

@@ -3,7 +3,7 @@ use std::future::pending;
 use std::sync::atomic::{AtomicBool, Ordering};
 use std::sync::{Arc, Mutex};
 
-use dashmap::DashSet;
+use clashmap::ClashSet;
 use redis::streams::{StreamReadOptions, StreamReadReply};
 use redis::{AsyncCommands, FromRedisValue, Value};
 use serde::Deserialize;
@@ -55,9 +55,9 @@ impl TryFrom<&Value> for ControlPlaneEvent {
 
 pub struct EndpointsCache {
     config: EndpointCacheConfig,
-    endpoints: DashSet<EndpointIdInt>,
-    branches: DashSet<BranchIdInt>,
-    projects: DashSet<ProjectIdInt>,
+    endpoints: ClashSet<EndpointIdInt>,
+    branches: ClashSet<BranchIdInt>,
+    projects: ClashSet<ProjectIdInt>,
     ready: AtomicBool,
     limiter: Arc<Mutex<GlobalRateLimiter>>,
 }
@@ -69,9 +69,9 @@ impl EndpointsCache {
                 config.limiter_info.clone(),
             ))),
             config,
-            endpoints: DashSet::new(),
-            branches: DashSet::new(),
-            projects: DashSet::new(),
+            endpoints: ClashSet::new(),
+            branches: ClashSet::new(),
+            projects: ClashSet::new(),
             ready: AtomicBool::new(false),
         }
     }

proxy/src/cache/project_info.rs

@@ -5,7 +5,7 @@ use std::sync::Arc;
 use std::time::Duration;
 
 use async_trait::async_trait;
-use dashmap::DashMap;
+use clashmap::ClashMap;
 use rand::{thread_rng, Rng};
 use smol_str::SmolStr;
 use tokio::sync::Mutex;
@@ -108,9 +108,9 @@ impl EndpointInfo {
 /// One may ask, why the data is stored per project, when on the user request there is only data about the endpoint available?
 /// On the cplane side updates are done per project (or per branch), so it's easier to invalidate the whole project cache.
 pub struct ProjectInfoCacheImpl {
-    cache: DashMap<EndpointIdInt, EndpointInfo>,
+    cache: ClashMap<EndpointIdInt, EndpointInfo>,
 
-    project2ep: DashMap<ProjectIdInt, HashSet<EndpointIdInt>>,
+    project2ep: ClashMap<ProjectIdInt, HashSet<EndpointIdInt>>,
     config: ProjectInfoCacheOptions,
 
     start_time: Instant,
@@ -176,8 +176,8 @@ impl ProjectInfoCache for ProjectInfoCacheImpl {
 impl ProjectInfoCacheImpl {
     pub(crate) fn new(config: ProjectInfoCacheOptions) -> Self {
         Self {
-            cache: DashMap::new(),
-            project2ep: DashMap::new(),
+            cache: ClashMap::new(),
+            project2ep: ClashMap::new(),
             config,
             ttl_disabled_since_us: AtomicU64::new(u64::MAX),
             start_time: Instant::now(),
@@ -302,7 +302,7 @@ impl ProjectInfoCacheImpl {
         let mut removed = 0;
         let shard = self.project2ep.shards()[shard].write();
         for (_, endpoints) in shard.iter() {
-            for endpoint in endpoints.get() {
+            for endpoint in endpoints {
                 self.cache.remove(endpoint);
                 removed += 1;
             }