safekeeper: don't allocate send buffers on stack (#9644)

## Problem

While experimenting with `MAX_SEND_SIZE` for benchmarking, I saw stack
overflows when increasing it to 1 MB. Turns out a few buffers of this
size are stack-allocated rather than heap-allocated. Even at the default
128 KB size, that's a bit large to allocate on the stack.

## Summary of changes

Heap-allocate buffers of size `MAX_SEND_SIZE`.

safekeeper/src/copy_timeline.rs

@@ -172,7 +172,7 @@ async fn copy_disk_segments(
 ) -> Result<()> {
     let mut wal_reader = tli.get_walreader(start_lsn).await?;
 
-    let mut buf = [0u8; MAX_SEND_SIZE];
+    let mut buf = vec![0u8; MAX_SEND_SIZE];
 
     let first_segment = start_lsn.segment_number(wal_seg_size);
     let last_segment = end_lsn.segment_number(wal_seg_size);

safekeeper/src/debug_dump.rs

@@ -383,7 +383,7 @@ pub async fn calculate_digest(
     let mut wal_reader = tli.get_walreader(request.from_lsn).await?;
 
     let mut hasher = Sha256::new();
-    let mut buf = [0u8; MAX_SEND_SIZE];
+    let mut buf = vec![0u8; MAX_SEND_SIZE];
 
     let mut bytes_left = (request.until_lsn.0 - request.from_lsn.0) as usize;
     while bytes_left > 0 {

safekeeper/src/send_wal.rs

@@ -467,7 +467,7 @@ impl SafekeeperPostgresHandler {
             end_watch,
             ws_guard: ws_guard.clone(),
             wal_reader,
-            send_buf: [0; MAX_SEND_SIZE],
+            send_buf: vec![0u8; MAX_SEND_SIZE],
         };
         let mut reply_reader = ReplyReader {
             reader,
@@ -548,7 +548,7 @@ struct WalSender<'a, IO> {
     ws_guard: Arc<WalSenderGuard>,
     wal_reader: WalReader,
     // buffer for readling WAL into to send it
-    send_buf: [u8; MAX_SEND_SIZE],
+    send_buf: Vec<u8>,
 }
 
 const POLL_STATE_TIMEOUT: Duration = Duration::from_secs(1);