Constrain MDF extraction streams to avoid pooled buffer bleed

[Copilot]

MDF extraction was writing more bytes than intended because ArrayPool rentals can exceed the requested size, leading to trailing garbage or data overlap in unpacked files. This change ensures only the actual payload length is exposed when writing.

• Pooled buffer safety
  • Wrap ArrayPool-backed buffers in length-bounded, read-only MemoryStream instances for MDF fast-path extraction.
  • Track the true decompressed size and prevent writing beyond it.

Example (length-bounded stream over pooled buffer):

mdfDecompressedLength = ms.MdfGetOriginalLength();
mdfDecompressed = ArrayPool<byte>.Shared.Rent(mdfDecompressedLength);
MdfShell.ToPsb(ms, mdfDecompressed, bodyContext);
mms = new MemoryStream(mdfDecompressed, 0, mdfDecompressedLength, false);

Original prompt

---

This section details on the original issue you should resolve

<issue_title>[Bug] ArrayPool causes garbage and overlapping data at the end of the file</issue_title>
<issue_description>Had an issues with script files, they were unpacking with parts of other scripts or bunch of null bytes. A friend of mine took a look and that's what he found.

At least for this bit, but you probably need to check other instances:

mdfDecompressed = ArrayPool<byte>.Shared.Rent(ms.MdfGetOriginalLength());

A length passed to the Rent is minimumLength, so your actual buffer migth be bigger and later you're writing all avaliable bytes from it
+ sometimes you might get the same array from the pool which will contain data from your previous file

</issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes [Bug] ArrayPool causes garbage and overlapping data at the end of the file #160

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.