Is your feature request related to a problem or challenge? Please describe what you are trying to do.
In the #122 and #1356, I add new datatype(decimal) and propose a more clear coercion rule for expr system.
When I add decimal to the SUM and AVG function #1408 , I meet some problem.
It's about the returned data type.

For example:
https://github.com/apache/arrow-datafusion/blob/415c5e124af18a05500514f78604366d860dcf5a/datafusion/src/physical_plan/expressions/sum.rs#L49 for floating-point datatype, the pg use the double-precision(double/float64) as the returned data type, but the datafusion is different.

There may be other different rules in spark

Which rule should we follow? For example sum and avg?
@alamb @houqp

Describe the solution you'd like
A clear and concise description of what you want to happen.

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
Add any other context or screenshots about the feature request here.