Implement JVM UDFs for all date/time expressions

[andygrove]

What is the problem the feature request solves?

Once the Comet JVM UDF work is finalized, I would like to implement JVM UDFs for all date/time expressions and make those the default so that we have 100% Spark compatibility, but allow users to opt in to Rust versions if they are ok with the documented compatibility issues.

We'll need to figure out configs - see #4310 for that discussion.

Describe the potential solution

No response

Additional context

No response

[parthchandra]

Not saying we shouldn't do this, but I very much recommend that we try to implement fully native as much as possible.

[andygrove]

Not saying we shouldn't do this, but I very much recommend that we try to implement fully native as much as possible.

The challenge is that many of our native date/time expressions are currently incompatible with Spark and disabled by default.

https://datafusion.apache.org/comet/user-guide/latest/compatibility/expressions/datetime.html

Chrono has a number of known limitations as well, such as limited timezone support. Without a Spark-compatible date/time library to use, it seems pragmatic to implement JVM versions so that we can enable the expressions by default.