feat(span-first): Better span first estimates (#5936)

We were underestimating the size of spans with particularly bulky
attributes, and flushing the buffer too late as a result. Take attribute
size into account for estimating.

Author: sentrivana

sentry_sdk/_span_batcher.py

@@ -91,13 +91,21 @@ def add(self, span: "StreamedSpan") -> None:
     @staticmethod
     def _estimate_size(item: "StreamedSpan") -> int:
         # Rough estimate of serialized span size that's quick to compute.
-        # 210 is the rough size of the payload without attributes, and we
-        # estimate additional 70 bytes on top of that per attribute.
-        return 210 + 70 * len(item._attributes)
+        # 210 is the rough size of the payload without attributes, and then we
+        # estimate the attributes separately.
+        estimate = 210
+        for value in item._attributes.values():
+            estimate += 50
+
+            if isinstance(value, str):
+                estimate += len(value)
+            else:
+                estimate += len(str(value))
+
+        return estimate
 
     @staticmethod
     def _to_transport_format(item: "StreamedSpan") -> "Any":
-        # TODO[span-first]
         res: "dict[str, Any]" = {
             "trace_id": item.trace_id,
             "span_id": item.span_id,
@@ -126,7 +134,7 @@ def _flush(self) -> None:
                 return
 
             envelopes = []
-            for trace_id, spans in self._span_buffer.items():
+            for spans in self._span_buffer.values():
                 if spans:
                     dsc = spans[0]._dynamic_sampling_context()