This is a useful function for parsing the formatted files, so
move this from rinex.c into rktcmd.c and export it.

Improve the ANTEX support.

* Support PCV with azimuth. Was just NOAZI. Use linear interpolation
  between azimuth points, as well as the zenith. Satellite PCV still
  only supports NOAZI as that is most common, but all the support is
  there to extend this to vary with azimuth

* Support ZEN1, ZEN2, DZEN, and DAZI. The PHV size is also now
  variable so the smallest size array is allocated. There was a fixed
  azimuth length of 19. It now correctly handles variations of these
  parameters. Require some more memory management, to free these.

* Support the full range of frequencies, across systems. Was just GPS
  and NFREQ. The usage of the antmodel() etc functions needed to
  change. There were returning a NFREQ array with the offsets, but
  that did not work with multiple systems with different frequencies
  mapped to the frequency same indices - these functions are now
  single frequency functions.

* Support PCO and PCV interpolation between frequency entries, with
  heuristics to avoid extrapolation of noisy data, and avoid
  interpolation across frequency bands (which may well be implemented
  with separate antenna elements e.g. stacked elements).

* Add support for ANTEX 2 which avoids much redundancy. This required
  the PRN to SVN mapping in the SINEX satellite meta data.

* Add minimal SINEX satellite meta data support, just for the PRN to
  SVN mappings. Adding the file-satmetafile option.

* Support for the console apps.

* Support for the qt apps, adding the satellite meta data file to the
  GUI.

* Some windows app support, untested, and not yet the satellite meta
  data file for ANTEX 2 which needs adding to the GUI.

* Adding a new function antpcv(). Implement antmodel() using the
  existing antpco() and the this new function to avoid code duplication,
  with some repetition of code execution.

  This is intended to allow the caller to apply the receiver PCO first
  and then separately the PCV, for greater accuracy, just as these are
  currently separated for satellites.

Implement the automated antenna type selection along with the antenna
delta from the input stream, from an RTCM or raw stream, the config
options ant{1,2}-anttype="*".

Have the config options ant{1,2}-postype RTCM and raw automatically
set the position, and independently of the antenna type, and now
including for the rover when in fixed mode.

Previously only an antenna delta would be picked up as an offset to a
base position, including a RTCM height, and only when the position was
also being automatically set and only for the base and not the rover.

Also there was no separate 'raw' antenna automated position mode, this
is included under the 'rtcm' option.

Pass in the receiver index, as a 'base' argument, which is 0 for the
rover and 1 for the base, to allow the respective options to be
selected rather than just using the rover options even for the
base. Start making use of this: in rescode() this is now used to
select the respective SNR and SNR mask.

For postpos.c where a one based receiver number and zero base index
were both be used, fix an instance in antpos() of indexing into option
anttype[] with an one base index which would be OOB for a value of
2. Try to clean up this code in this regard.

This allows setpcv() to automatically set the antenna delta, rather
than having to hack that delta into the position in antpos(), and that
hack can now be removed.

Have tropmodel use the precise mapping function, rather than 1/cos(z).

Update the exponent in the pressure vs temperature function, from
5.2568 to 5.255877 which is found in references on the matter.

Update the rtkpos RTK code to also account for the wet component when
not estimating it, and to also allow use of the SBAS model when that
is selected.

Apply the antenna delta from the marker to ARP, and the per-frequency
PCO, before some of the other calculations such as the geometric
distance and line of sight vectors. The antenna PCV is later applied.
The line of sight vectors and now from the antenna phase center,
rather than from the marker position. This makes the calculations much
more robust to large antenna offsets. It was a trivial matter to also
use per-frequency line-of-sight vectors, so do that.

The azimuth and elevation are now also from the phase center rather
than the marker position, but still only per satellite.