make ApplyNameMapping keep field's name after name mapping.

[zhongyujiang]

This PR makes ApplyNameMapping keep field's name after name mapping, which is for compatibility while reading the migrated Iceberg table with legacy parquet schema mentioned in this issue #3604.

[chenjunjiedada]

Looks correct, could you please add a unit test for this？

[zhongyujiang]

Looks correct, could you please add a unit test for this？

Of course!

[zhongyujiang]

Hi @chenjunjiedada, I found that RemoveIds haven't keep the original name of repeatedElement of list/map type too:

@Override
  public Type list(GroupType array, Type item) {
    return Types.list(array.getRepetition())
        .element(item)
        .named(array.getName());
  }

I think this also need a fix.
And speaking of unit test, I found that TestPruneColumns already did a similar one for PruneColumns, maybe I can reuse the fileSchema in it, so can I just rename TestPruneColumns to TestParquetTypeVisitor and put the unit tests for RemoveIds and ApplyNameMapping in it? What do you think?

[zhongyujiang]

Hey @chenjunjiedada , I make RemoveIds keeps field's original name too and add unit tests for both RemoveIds and ApplyNameMapping in the latest commit, could you please give it a review?

[uncleGen]

In general, this pr makes sense to me, and is there same issue in ORC ff?

[uncleGen]

Besides, could you please add an end-to-end unit test?

[zhongyujiang]

In general, this pr makes sense to me, and is there same issue in ORC ff?

Sorry, I don't know much of ORC, and I think this issue is already solved by this pr, it's already merged. I'll close this one.