proxy: reduce global conn pool contention (#4747)

## Problem

As documented, the global connection pool will be high contention.

## Summary of changes

Use DashMap rather than Mutex<HashMap>.

Of note, DashMap currently uses a RwLock internally, but it's partially
sharded to reduce contention by a factor of N. We could potentially use
flurry which is a port of Java's concurrent hashmap, but I have no good
understanding of it's performance characteristics. Dashmap is at least
equivalent to hashmap but less contention.

See the read heavy benchmark to analyse our expected performance
<https://github.com/xacrimon/conc-map-bench#ready-heavy>

I also spoke with the developer of dashmap recently, and they are
working on porting the implementation to use concurrent HAMT FWIW

proxy/src/http/sql_over_http.rs

@@ -44,6 +44,7 @@ const MAX_REQUEST_SIZE: u64 = 1024 * 1024; // 1 MB
 
 static RAW_TEXT_OUTPUT: HeaderName = HeaderName::from_static("neon-raw-text-output");
 static ARRAY_MODE: HeaderName = HeaderName::from_static("neon-array-mode");
+static ALLOW_POOL: HeaderName = HeaderName::from_static("neon-pool-opt-in");
 static TXN_ISOLATION_LEVEL: HeaderName = HeaderName::from_static("neon-batch-isolation-level");
 static TXN_READ_ONLY: HeaderName = HeaderName::from_static("neon-batch-read-only");
 static TXN_DEFERRABLE: HeaderName = HeaderName::from_static("neon-batch-deferrable");
@@ -193,7 +194,7 @@ pub async fn handle(
     let array_mode = headers.get(&ARRAY_MODE) == Some(&HEADER_VALUE_TRUE);
 
     // Allow connection pooling only if explicitly requested
-    let allow_pool = false;
+    let allow_pool = headers.get(&ALLOW_POOL) == Some(&HEADER_VALUE_TRUE);
 
     // isolation level, read only and deferrable