Appends a chunk of data to an in-progress upload session. This is the recommended approach for large blobs, as each chunk is stored immediately rather than buffering the entire blob in memory.

Clients can call this endpoint multiple times to upload a blob in pieces, then call PUT /v2/{name}/blobs/uploads/{session_id}?digest=... to complete the upload.

The Content-Range header is optional but recommended - if provided, the server validates that the chunk starts at the expected offset.