Support Arrow IPC Big Endian / Cross Endian

[alamb]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.

The arrow-rs Arrow IPC implementation is tested thorougly on little endian platforms and works well 👍

However, it is not clear how well it will work on big endian systems and I don't think we have automated test coverage for that yet.

After #3427, it is quite clear that the arrow-rs support will not work well when the IPC stream is a different endianness to the host platform.

However I am not sure that it actually important to test and support cross endianness given the spec says
https://arrow.apache.org/docs/format/Columnar.html#byte-order-endianness

The main use case is exchanging RecordBatches between systems with the same Endianness. At first we will return an error when trying to read a Schema with an endianness that does not match the underlying system.

Currently the arrow-rs IPC reader does not attempt to convert endianess of the ipc streams, nor does it (always) error out when that endianess is different. It will error in some cases (like dictionaries #859 and Decimal arrays)

Describe the solution you'd like
I suggest throwing an error when source/target endianess is not the same to make it clear this is not supported

Then we can plan out such support if/when we ever want

The integration tests could offer a good starting point (there are bigendian files and the CI on github runs on little endian runners x86_64).

• https://github.com/apache/arrow-rs/blob/81abc1a942cd13a92231f4a828077ad60fdabe36/arrow-integration-testing/tests/ipc_writer.rs
• https://github.com/apache/arrow-rs/blob/81abc1a942cd13a92231f4a828077ad60fdabe36/arrow-integration-testing/tests/ipc_reader.rs

Describe alternatives you've considered
N/A
Additional context
Found while working on #3427