fix: support decoding application with duplicate res entries

[iBotPeaches]

fixes: #2824

In the sample above. We find a duplicate resource at end of chunk.

      [ResTable_entry] id: 0x0595 name: zero_corner_chip keyIndex: 5176 size: 8 flags: 0x0004
        [Res_value] size: 8 dataType: 0x12 data: 0x00000000 ((id))
      [ResTable_entry] id: 0x0596 name: zoom keyIndex: 5177 size: 8 flags: 0x0004
        [Res_value] size: 8 dataType: 0x12 data: 0x00000000 ((id))
      [ResTable_entry] id: 0x0597 name: zoom keyIndex: 5177 size: 8 flags: 0x0004
        [Res_value] size: 3 dataType: 0x12 data: 0x00000000 ((id))

The last entry has the same index to the previous and malformed. The size reported is also wrong. There are still 8 bytes, but its a duplicate entry.

I've debated a few paths.

• Detecting read vs actual during value read
• Skipping reading if we have less than 8 bytes in value read

In this patch we return early from value read if size less than 8, fallback to null and let the end chunk reading (read vs actual) skip what is remaining.