XM data corruption

[KirillKryukov]

XM compressor still has data corruption issue. Compressing some input and decompressing it back produces corrupted output. I.e., decompressed data is different from original file.

Test data size: 30,244 bytes
Test data link: http://kirill.med.u-tokai.ac.jp/data/temp/xm-repro-4-input.zip

Commands to reproduce:

Compress:
jsa.xm.compress --hashSize=11 --context=15 --limit=200 --threshold=0.15 --chance=20 --real=archive.xm original.fasta

Decompress:
jsa.xm.compress --hashSize=11 --context=15 --limit=200 --threshold=0.15 --chance=20 --decode=archive.xm --output=decompressed.fasta

Compare:
cmp original.fasta decompressed.fasta

Produces: original.fasta decompressed.fasta differ: byte 27512, line 274

The decompressed file has correct size, but corrupted sequence data. It was found during testing for Sequence Compression Benchmark.

Let me know if you need any additional information or help.