- following Andy's idea to return values as label whenever possible
      eg, 1.2e6 -> 1200000
  but left it commented out

- avoid buffer overflow in ISstream::read(word&).
  Is the 'if (fail())' check itself actually in the correct place??

- other minor cosmetic changes

- since atof is just a call to strtod (in glibc at least), revert to the
  original character collection, convert the number via strtod/strtol and
  check the end pointer so see if everything was converted. Flag both bad
  formats (eg, ".e") and trailing junk (eg, "1.e4e") as a tokenize error.
  These were previously silently ignored.

- simple (same weight as before), but with trivial error checking.
  Retain buffer overflow check.

- allow construct with Xfer container for the addressing

- Replaced non-const addressing() method in BiIndirectList with
  resetAddressing() method as per IndirectList

- only ignore html files in the main directory, not everywhere.
  This is how we lost FoamFooter.html, FoamHeader.html in doc/Doxygen/ in
  the 1.6.x release

- read ~OpenFOAM/foamCopySettings.rc file for an extra rsync options
  for example, to ignore post-processing or sampling directories

- avoid other degenerate sequences from being treated as a scalar
  eg, '1-e', '1e+', '1e.2'

- for alphanumeric sequences (optionally with [-+.]) that don't look
  like a float or int -> return as a word

  This means that '0patch' now looks like a <word> (not <label>)
  and '1.end' looks like a <word> and not <scalar>

  Something like '1-e' still does get treated as a bad <scalar> though

- require a digit somewhere. This traps '.' or '.+E' type of sequences
  instead of returning them as 0

- avoid buffer overruns on really long sequences

- should we be considering strtod() as an alternative to atof() and checking
  the errno?
  The same might be true for using strtoul() or strtol() instead of atol()

- solves problems that can occur when checking $PWD and links are involved