I see no other solution, too. When I used this program for genome assembly I expected it to require lots of memory since it is normal to have millions of reads to assemble from. But here, using as small data sets as possible, I am surprised to find that I cannot modify program's desire to preallocate memory.
I have modified run-unit-test so that it runs for *-31mer* and *-63mer* binaries now. I have also fixed some lintian warnings and pointed debhelper version to 11. Please, check my commits!
With regards,
Liubov