Interesting behavior of the compiler

Hi Manuel,

It is again the time for Bigloo hacking. I have uncovered a possible 
potential divergent, or exponential computation in the compiler, i.e.,
it hangs in the "Globalization" pass on the test-1.scm (attached).

For comparison I also include test-2.scm which works with the native 
back-end (all options I tried) and with the JVM (only with -O, not
higher optimization options such as -O2), which BTW is just as fast as
the natively compiled version. The scripts exercise the same features
of Bigloo and miniKanren, but I have tried to isolate divergence in the 
compiler to test-1.scm.

Both scripts work just fine in the interpreter (only 10x slower of 
course). They should exit without printing (test-2.scm prints just one 
newline), and make extensive use of the SLIB and miniKanren ports (both 
attached as x86-64 binaries).

But I have to admit I am using 4.0b on an older (Ubuntu 12.04) Linux. 
Should have some time soon to upgrade.

Merry Christmas and Happy New Year!

Kind regards,
Peter Kourzanov

P.S. Hope to see you in Delft in January;-)