fix: remove delay in publish_batch execution

The existing code creates futures for publish_batch with delays to
spread the rpc load evenly in the publish interval however due to the
Rust async design those futures do not get started unless they are
awaited. This results in additional deterministic latency of almost the
publish_interval especially for the publishers who publish many prices.

This change removes the code as even the intended behaviour adds latency
and disrupts the publish interval (the last publish_batch can block the
actor loop).

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "pyth-agent"
-version = "2.5.0"
+version = "2.5.1"
 edition = "2021"
 
 [[bin]]

src/agent/solana/exporter.rs

@@ -527,23 +527,15 @@ impl Exporter {
         // Split the updates up into batches
         let batches = permissioned_updates.chunks(self.config.max_batch_size);
 
-        // Publish all the batches, staggering the requests over the publish interval
-        let num_batches = batches.len();
-        let mut batch_send_interval = time::interval(
-            self.config
-                .publish_interval_duration
-                .div_f64((num_batches + 1) as f64), // +1 to give enough time for the last batch
-        );
         let mut batch_state = HashMap::new();
         let mut batch_futures = vec![];
+
         for batch in batches {
             batch_futures.push(self.publish_batch(batch));
 
             for (identifier, info) in batch {
                 batch_state.insert(*identifier, (*info).clone());
             }
-
-            batch_send_interval.tick().await;
         }
 
         // Wait for all the update requests to complete. Note that this doesn't wait for the