Fix Compress::Gzip::Header to check for CRC16

[kojix2]

I wrote code to check the value of CRC16 if the Gzip header flag contains an HCRC.

This solves part of #14789

[kojix2]

Feel free to correct me if you have a better way to write this.

[straight-shoota]

As always, this change is void if there's no spec to ensure its functionality. We'll need at least one example with an invalid CRC and a test to ensure it raises.

Capturing the header bytes in an array seems very inefficient. It also shouldn't be necessary to do in one go.
Instead of passing the entire data to CRC32.checksum, we can update the checksum with individual bytes or slices as we read them. The result should be identical.
We can either instantiate the CRC32 class and append data bits, or use the .initial and .update class methods. The latter is used in the Gzip::Reader and Gzip::Writer implementations, for example.

[ysbaddaden]

Suggestion: what about computing the CRC value as we parse the data instead of allocating many objects into the HEAP?

For example you can check flg.hcrc? first to create a digest object, then check if it exits to update the checksum.

if flg.hcrc?
  crc = Digest::CRC32.new
  crc.update header.to_slice
end

if flg.extra?
  # ...

  if crc
    crc.update(xlen.unsafe_as(...).to_slice
    crc.update(@extra)
  end
end

# ...

if crc
  checksum = 0_u32
  crc.final(Slice.new(pointerof(checksum).as(UInt8*), 4)

  crc16 = io.read_bytes(UInt16, ...)
  raise "mismatch" unless checksum == crc16
end