Incorrect input schema when preparing result expressions for HashAggregation #502
Description
Describe the bug
If there are result expressions for Spark HashAggregation, we prepare these physical expressions to add an additional DataFusion ProjectionExec on top of AggregateExec.
Currently we incorrectly use AggregateExec's child schema as input schema to prepare these physical expressions. It causes some test failures in #403 as latest DataFusion adds additional check for that in apache/datafusion#10088.
Steps to reproduce
No response
Expected behavior
No response
Additional context
No response