OpenBSM 1.0

- Import of Darwin74 BSM drop
- Use 'syslog' for audit log warnings, rather than echoing to a file in
  audit_warn.
- Compile using BSD make infrastructure.
- Integrate bsm/ include files from Darwin74 XNU drop into OpenBSM.
- Narrow set of symbols and defines that are exposed in user space: don't
  compile in code relying on kernel-only types such as 'struct socket'.
- Add README, including basic build documentation.
- Compilation of Apple-specific notify and Machroutines now #ifdef __APPLE__.
- Staticize libbsm global variables to avoid leakage into application.
- Add free_au_user_ent() so that au_user_ent's don't have to be leaked.
- Clean up bogus nul-termination checks in libbsm.
- Add libbsm API man pages: au_class.3 au_control.3 au_event.3
  au_free_token.3 au_io.3 au_mask.3 au_token.3 au_user.3 libbsm.3.
- Add man pages for BSM system calls: audit.2 auditctl.2 auditon.2 getaudit.2
  getauid.2 setaudit.2 setauid.2
- Modify various libbsm interfaces to more consistently return 'errno' values
  on failure.
- Break out au_close() into constituent parts, allowing records to be written
  to memory as well as files.
- Prefix various defines with 'BSM_' to reduce name space pollution.
- Added audit_internal.h, which can be used by a kernel audit implementation
  wanting to rely on libbsm components.
- Build with warnings, and eliminate warnings.
- Make libbsm endian-independent, storing and reading BSM are big endian
  (network byte order) rather than native byte order.  More consistently
  print IP addresses using the IP address print routine.  These changes
  make use of sys/endian.h from *BSD; since this isn't present on Darwin,
  add it to OpenBSM as compat/endian.h, which is used only on Darwin.
- Import of Darwin80 BSM drop, including 64-bit file IDs, better
  documentation of private APIs, and bug fixes.
- White space cleanup.
- Add audit.log.5, a first cut at a man page documenting the BSM file format.
- Teach au_read_rec() to recognize stand-alone file tokens, which are present
  at the beginning and end of Solaris audit trails.  Technically, these
  appear to violate the high level BSM spec, which suggests that all tokens
  are present in records, but need to be supported.
- Implement HEADER64, ATTR64, SUBJECT64 token types, which make it possible
  to run praudit(1) on basic Solaris BSM streams.
- Switched to Solaris spelling of token names; Darwin spellings are now
  deprecated and will be removed in a future version of OpenBSM.
