better to use duration than time

dev/restate/service/protocol.proto

@@ -56,10 +56,9 @@ message StartMessage {
   // thus it's susceptible to Restate's crashes/leader election changes.
   uint32 retry_count_since_last_stored_entry = 7;
 
-  // Duration since the last stored entry.
-  // The time is set as duration since UNIX Epoch in milliseconds.
+  // Duration since the last stored entry, in milliseconds.
   //
-  // Please note this time might not be accurate,
+  // Please note this duration might not be accurate,
   // and might change depending on which Restate replica executes the request.
   uint64 duration_since_last_stored_entry = 8;
 }
@@ -106,10 +105,9 @@ message ErrorMessage {
   // Entry type.
   optional uint32 related_entry_type = 6;
 
-  // When to execute the next retry.
-  // If provided, it will override the default retry policy used by Restate's invoker.
-  // The time is set as duration since UNIX Epoch in milliseconds.
-  optional uint64 next_retry_at = 8;
+  // Interval before executing the next retry, specified as duration in milliseconds.
+  // If provided, it will override the default retry policy used by Restate's invoker ONLY for the next retry attempt.
+  optional uint64 next_retry_interval = 8;
 }
 
 // Type: 0x0000 + 4