Binary decoding likely has 32bit issues with large data sizes

[thomasvl]

Like what #1375 fixed, it looks like we've got a few cases (getFieldBodyBytes(count:), consume(length:), ...) where because the internals of the BinaryDecoder use Int we could have failures in a 32bit build.

[thomasvl]

fyi - The C++ upstream code seem to use INT_MAX in some cased to catch limits, but it seems like that value will change depending on a build being for a 64 or 32 bit platform. And the new https://protobuf.dev/programming-guides/encoding/#cheat-sheet section calls out some 2GB limits on specific things.

[thomasvl]

Correction on that comment - int (and INT_MAX) is still 32bit on most 64bit systems: https://en.wikipedia.org/wiki/64-bit_computing#64-bit_data_models

[tbkka]

If there's a 2GB limit, we should still write it out as 0xffffffff or similar.

[thomasvl]

I'm going to go ahead and close this, I doing another pass, and don't immediately see issues, so at this point I'd suggest we treat them as bugs if found instead of keeping this open longer term without any clear next steps.