Header:
- 4 bytes <- number of files
For each entry:
- 4 bytes <- unknown, maybe a checksum or hash
- 4 bytes <- file offset (aligned to 4 bytes)
- 4 bytes <- file length unpacked/uncompressed
- 4 bytes <- number of file parts
Each file has a header with offsets to each part (the last offset is the offset to the next file) and the data packed or compressed (no idea).