fix: change map nullable value to false

[cmackenzie1]

Description

This value was true but where arrow defines it as always false https://github.com/apache/arrow-rs/blob/master/arrow-schema/src/field.rs#L230.

This is also described in apache/arrow-rs#1697.

This also replaces key_value as the struct name with entries to remain consistent with https://github.com/apache/arrow-rs/blob/878217b9e330b4f1ed13e798a214ea11fbeb2bbb/arrow-schema/src/datatype.rs#L247-L250

The description of the main changes of your pull request

Related Issue(s)

• closes Internal Parquet panic when using a Map type.  #1619

Documentation

[wjones127]

This makes sense. Both entries and keys are required to be non-nullable by the Arrow spec :)

https://github.com/apache/arrow/blob/c4b01c60fba85bbfc3a1b1510e179153c0f79515/format/Schema.fbs#L124

[wjones127]

I think we are hitting an upstream bug in MapArray::new_from_strings(): apache/arrow-rs#4807

[wjones127]

@cmackenzie1 for the test_record_batch_from_map_type error, I think we should regard MapArray::new_from_strings() as broken and change the test to build a map array differently for now.

I'm unclear on what the parquet2 code is doing near the other failure; I haven't touched that code base yet. LMK if you want more eyes on that and I can dig in.

[cmackenzie1]

@wjones127 I reverted the change to parquet2 - not sure what is up with that one. Also I yanked the MapArray::new_from_strings and applied the patch inline for now. We can clean that up once arrow is upgraded here.