docs: document schema evolution behavior for native_datafusion scan

Author: andygrove

docs/source/contributor-guide/parquet_scans.md

@@ -62,6 +62,10 @@ cause Comet to fall back to Spark.
 - No support for `input_file_name()`, `input_file_block_start()`, or `input_file_block_length()` SQL functions.
   The `native_datafusion` scan does not use Spark's `FileScanRDD`, so these functions cannot populate their values.
 - No support for `ignoreMissingFiles` or `ignoreCorruptFiles` being set to `true`
+- Schema evolution (type promotion such as int→long, float→double) is always enabled. DataFusion's native Parquet
+  reader handles type promotion automatically, regardless of the `spark.comet.schemaEvolution.enabled` setting.
+  This means queries that would fail in Spark or other Comet scan implementations when schema evolution is disabled
+  will succeed with `native_datafusion`.
 
 The `native_iceberg_compat` scan has the following additional limitation that may produce incorrect results
 without falling back to Spark: