[EPIC] Schema metadata handling / bugs

[alamb]

Is your feature request related to a problem or challenge?

As described by @wiedld on https://github.com/apache/datafusion/pull/12729

A https://github.com/apache/datafusion/pull/11989 (and released in 42.0.0) has added a check that the logical and physical plan input schemas (to an aggregate node) are the same.

This check fails for us in several different queries now downstream in InfluxDB and this ticket tracks the list of issues we found

Issues

• [x] https://github.com/apache/datafusion/issues/12687
• [x] #12734
• [x] #12735
• [ ] https://github.com/apache/datafusion/issues/12736
• [x] https://github.com/apache/datafusion/pull/12951
• [ ] https://github.com/apache/datafusion/pull/12954

[wiedld]

This check fails for us in several different queries now downstream in InfluxDB and this ticket tracks the list of issues we found

If anyone finds other bugs, feel free to add to the epic. 🙏🏼

[alamb]

@wiedld has found several more issues:

• [x] https://github.com/apache/datafusion/pull/12951
• [x] https://github.com/apache/datafusion/pull/12954

[eejbyfeldt]

Found another issue that triggers a schema issue here: https://github.com/apache/datafusion/issues/13190

[alamb]

Found another issue that triggers a schema issue here: #13190

Thanks -- added to the list

Note I now have added the config option to work around the issue in https://github.com/apache/datafusion/pull/13176 for anyone who hits these

[wiedld]

Found another schema issue here: https://github.com/apache/datafusion/pull/13305. Reproducer test cases are also included.

[alamb]

The subtasks are largely done , let's call this one done

[adriangb]

Found another one: https://github.com/apache/datafusion/pull/19335