Prepared for parsing the new data formats. Data isn't formatted
or parsed yet, though.

No more memory leaks! Unfortunately each memory leak was replaced
by a potential buffer overflow. Basically, the code was riddled
with functions that accept a char ** data, which
is malloced and then formatted. This was then leaked.
Now, it accepts a char * data, which is allocated on the stack
by the calling function. However, there is no bounds checking,
because that is slightly more involved.

See todos in format_command and parse_packet.

Untested

Was done by copying files, rather than a merge. This nuked the
commit history.