[glTF 2.1] 64-bit upgrade to binary format

# 64-bit upgrade to the binary format

glTF 2.0's binary format (`.glb`) uses 32-bit integers for lengths. glTF 2.1 provides a upgraded binary format that uses 64-bit integers for lengths. This is known as binary format version 3, and the existing binary format (version 2) is still defined.

Additionally, a new chunk encoding field is added to the chunk header for version 3 only. For glTF 2.1, this field is always zero. Future versions of glTF or extensions may use this field.

Importers must support both binary format version 2 and version 3. Exporters may choose to keep using version 2 binary format for compatibility, so long as the files are smaller than 4 GiB and do not use the new chunk encoding field.

## Use cases

The 64-bit binary format allows for files larger than 4 GiB. The upgraded limit is 8 EiB, with the most significant bit reserved for future use.

The chunk encoding field may be used in the future for chunk-level compression, encryption, or other features. glTF 2.1 does not define any of those, however glTF extensions already allow for compression at other levels.

## Example

> [!NOTE]
> Examples are illustrative and not exhaustive.

## Why this design

The 8 EiB limit ensures that the most significant bit is always zero, allowing glTF binary parsers to safely use signed 64-bit integers, as well as unsigned 64-bit integers. This allows for easier implementation in signed-only languages, allows `-1` to be used as a sentinel or error value, and reserves the most significant bit for future use if needed.

Start by reviewing the existing glTF 2.0 binary format definition alongside the 64-bit version 3 requirements described here. The issue's design must be resolved and documented so importers support versions 2 and 3 while exporters retain the stated compatibility rules.

Written by the indexing model from the issue text.