docs(sdk): update log attribute dedup benchmark numbers

Align bench header with latest local criterion run for PR #3537.

Author: Anant Kumar

opentelemetry-sdk/benches/log_record_attribute_dedup.rs

@@ -8,14 +8,14 @@
 //! Results (unique keys — typical case, no duplicates in the batch):
 //! | Scenario                | Dedup ON | Dedup OFF (before) | Overhead |
 //! |-------------------------|----------|--------------------|----------|
-//! | add_1_unique_attribute  | 25.6 ns  | 25.9 ns            |  ~1.0x   |
-//! | add_5_unique_attributes | 129.3 ns | 127.5 ns           |  ~1.0x   |
-//! | add_9_unique_attributes | 292.0 ns | 223.6 ns           |  ~1.3x   |
+//! | add_1_unique_attribute  | 27.2 ns  | 24.7 ns            |  ~1.1x   |
+//! | add_5_unique_attributes | 135.5 ns | 119.3 ns           |  ~1.1x   |
+//! | add_9_unique_attributes | 301.2 ns | 220.8 ns           |  ~1.4x   |
 //!
 //! Results (repeated key — duplicate writes to the same key):
 //! | Scenario       | Dedup ON | Dedup OFF (before) |
 //! |----------------|----------|--------------------|
-//! | add_5_same_key | 32.1 ns  | 38.5 ns            |
+//! | add_5_same_key | 31.9 ns  | 32.1 ns            |
 //!
 //! Note: criterion does not fail CI on regression by itself. These numbers are
 //! reference values for human review in PR #3537 / issue #3497.