[ntuple] Properly support big-endian architectures

[jalopezg-git]

Explain what you would like to see improved

RNTuple metadata is serialized to a machine-independent representation. However, contents of a page are written using the native endianess of the architecture on which RNTuple runs. According to the RNTuple v1 specification, these should be laid out always in little-endian order.

Optional: share how it could be improved

If running on a big-endian machine, (un)packing of data pages for some data types should carry out a (16, 32, or 64-bit) byte swap, depending on the type.

[GitHub Actions]

Hi @jalopezg-r00t,

It appears this issue is closed, but wasn't yet added to a project. Please add upcoming versions that will include the fix, or 'not applicable' otherwise.

Sincerely,
🤖