useless statistics on the current ports tree

[Marc Espie <espie_(_at_)_schutzenberger_(_dot_)_liafa_(_dot_)_jussieu_(_dot_)_fr>]

A build against a -stable machine gives the following results.

Marked as broken:
gom-0.29.103
xmp-1.1.5
hylafax-4.0.2
mgetty-1.1.21
tkhylafax-3.0
fxtv-1.02
kaffe-1.0.5 (problem with the new gmp)
libgcj-2.95 (obsolete)
mh-6.8.4
pgpsendmail-1.4.5
mc-4.1.35
cnews-cr.g
LPRng-3.6.12

Plain broken:
bochs-2000_0325a-debug (missing protos, deadly for C++)
imagemagick-5.2.9 (as usual, distfile  changed)
gphoto-0.4.3 (depending on imagemagick)
xfmail-1.4.6 (bad checksum)
gacc-0.7.5
impress-1.1b6
pgplib
uvscan_dat (missing distfile)
wwwcount-2.5 (still, bad checksum)

Not built:
gtl-1.0 (fetching source has restrictions)
cassowary-0.60 (depending on gtl-1.0)

Unfaked ports (hence not built):
prc-tools-0.5.0
wordperfect-8.0
wine-990225 (antique, anyway)
erlang-47.4.0
exmh-2.0.2
amanda-2.4.1p1
lyx-1.0.0
pkfonts300-1.0
webmin-0.81
nscache-0.2p1

There were a few more errors which turned out to be trivial to fix.
Most of these problems are transient. However it would be VERY cool
if the following ports could finally get fixed (yes, I consider non-faked
ports to be a bug, more often than not):
wwwcount-2.5
erlang-47.4.0
amanda-2.4.1p1
lyx-1.0.0 (note that lyx is positively ancient).
gom-0.29.103
hylafax-4.0.2
mgetty-1.1.21
LPRng-3.6.12
mc-4.1.35
pgpsendmail-1.4.5

as these HAVE had problems basically *forever*.

Now for the statistics. I end up with 1090 packages being built, out
of which 1039 are suitable for ftp.
The ftp-able packages take up 553MB, roughly. This would still fit on a
single CD, we still have 100MB to go.

On the other hand, the distfiles no longer fit on a CD: total is 943MB.

A full ports build tree for a 386 (with packages, distfiles, and 
working directories) roughly takes 9.5GB.