Many thanks for this detailed problem report.
I believe that's an unintended (and surprising) problem, it should still work. The trouble is, it works on all my computers (I don't have a 32-bit arch). So it's hard for me to figure out.
You can maybe create an i386 chroot in an amd64 system, and test stuff there - this is how I'm testing it as well.
If you replace your command with:
$lastal -vvv humdb fuguMito.fa
(verbose mode), what does it print before hanging?
$lastal -vvv humdb fuguMito.fa
/usr/lib/last-align/lastal-plain: batch size=8192
/usr/lib/last-align/lastal-plain: calculating matrix probabilities...
/usr/lib/last-align/lastal-plain: matrix lambda=1.09861
/usr/lib/last-align/lastal-plain: matrix letter frequencies (upper=reference, lower=query):
A C G T
25 25 25 25
25 25 25 25
/usr/lib/last-align/lastal-plain: getting E-value parameters...
1.0960171987681839, 0.33538787507026158,
2.0290734315292083, -0.46514786408422282,
2.0290734315292083, -0.46514786408422282,
5.0543294182155085, 15.130999712620039,
5.0543294182155085, 15.130999712620039,
5.0543962679167036, 15.129930117400917
Hangs after this bit.
It would be great if you could test versions between 1145 and 1167 (e.g. by "hg update -r1155"). Of course I can't expect you to go to that trouble.
I tried it for version 1155, and it times out with the exact same log I pasted above for 1167. Which means something broke between 1145 -1155.
Admittedly, I did not inspect any further.
Please let me know if you'd need any more information.
Thanks and regards
Nilesh