--* version 1.0.4 *--
Apperently I need to link to libm to use the pow() function. Don't ask why it
worked before, without it...

--* version 1.0.3 *--
Code was cleaned up to produce less warnings and prevent possible crashes,
as well as documentation improvements.

--* version 1.0.2 *--
-s flag was added to allow user to specify size of output file. If it's smaller
than the smallest input, it will be clipped to match, if larger, random data
will be appended.

--* version 1.0.1 *--
A limit was placed on the maximum number of (input files + random pads) in order
to prevent possible segfaults if a user tried more and exceeded the memory
limit. This will not be needed after the file pointer arrays can expand to meet
memory needs. A few more comments were added to the code, for good karma, and
a couple small cleanups.

--* version 1.0 *--
Code completely rewritten. Now supports any number of input files, and as many
random pads as the user specifies. Code is much cleaner and good-looking.

--* version 0.7 *--
Output files are now named according to David Madore's sceme, involving the
md5-sum of the data.

--* version 0.6 *--
Now works on FreeBSD (tested on 4.0-STABLE); other systems should also work now.
Code cleanups and documentation improvements.

--* version 0.5 *--
Pad now uses OpenSSL as a source for random bytes. This should increase
security, especially on operating systems that use a linear-type alorithm to
implement /dev/urandom.

--* version 0.4 *--
No longer forces output to be written to a file. It can write to stdout, which
is useful when being called from a shell script or other program. Code cleanups.

--* version 0.3 *--
Code cleanups and minor bugfixes.