Finally, the AllSky data generator is in good shape. It properly caches the previous valid cloudiness value for each station, a bug in the cloudiness calculation has been found (for low sun angles)and the documentation is now in order. We just need to get a cloudiness parameter in MeteoData in order to also be able to read measured cloudiness!

More improvements to the AllSky data generator: it now computes the cloudiness if none is provided. Another improvement will have to be brought for working with multiple stations as well as adding the cloud fraction to the MeteoData parameters! (ie: stay tuned!)

Fix for issue 366: DEMObject buffering, it was buffered in a map as Grid2DObject, now in a vector as DEMObject.

Added more flexibility for 2D resampling: it is now possible to resample an Array2D with different x and y factors. Improved the hillshade call in DEMObject. 

Since the number of clear or all sky parametrizations seem to multiply, two generic generators have been implemented: clearsky and allsky that take as argument which parametrization to use. Therefore the old Brutsaert or Dilley have been removed. The Unsworth should follow at some point, after we will have added the cloud fraction as a meteo parameter and found a way to also specifiy a parametrization for clear_sky_index <-> cloudiness that could be used by the all sky generators.

Moving code around in Atmosphere (to keep the clear sky parametrizations together).

Added a Dilley clear sky generator. The Unsworth does not provide clear sky as a fallback anymore, this has to be configured by the user (so the user can choose which fallback to enable).

small fix to support slightly different naming schemes between IMIS and MeteoCH (with an optional switch to enable it)

Fixed a segfault when using PNGIO (allocating the wrong number of elements). Small cleanup in Graphics.

PSQLIO: Adding explicit copy constructor and assignment operator to get rid of warnings regarding the psql pointer.

[INPUT]
METEO = PSQL
EXCLUDE = exclude.txt

The exclude file has the structure:
stationID, RH, TA, ...

The parameters will be shadowed and set to IOUtils::nodata

Code cleanup and adjusting to the MeteoSwiss naming scheme as well as VW_MAX vs VW trick (VW is Kalman-corrected and therefore can get higher than VW_MAX).

The COSMOXMLIO documentation has been updated and the plugin can now read all files in a given directory (but they must follow a loose naming scheme). Otherwise, small fixes (documentation typo, constification).

First version of the new COSMO (in fact, FieldExtract) plugin. It currently only reads data from one single file that is provided as argument, but this file can contain all the stations and all the time steps. This plugin only relies on libxml2.

Finally a good documentation on ordinary kriging... This should be a reasonnably standard (since everybody seems to enjoy using his own notation for the various matrices) and more importantly, really show what the whole thing is about, how the matrices are filled and what is solved.

Removing some try/catch, adding some documentation and adding the missing error message as reported by issue 357

Fixing a numerical bug: it happended once (Today) that the calculated cosinus of the solar azimuth angle was slightly less than -1... This lead to an arithmetic exception. Now, all values are brought back within [-1, +1] before computing the acos.

Removing the passing of ostrinstream objects by reference, since this forces us to reset the stringstream back to its original state and this lead to some compilation issues on some macs. Now, the interpolation messages are passed as normal strings.

Preparing the way for using IDW with an exponent on the distance. This is currently hardcoded to 1 (bypassing the call to "fast_pow" if it is 1 in order to have no performance impact in this case). We still have to figure out how the user could provide this value as an argument... In terms of performance, it is around 3% slower when calling fast_pow(x, 1.) compared to skipping this call. On some other tests, it was more on the order of 8%.

Preparing the way for altitude thresholds on lapse rates: for example, preventing HNW to get crazy at elevations much higher than the highest station. This would be achieved by using max_altitude instead of the true elevation of a high point when doing detrending/retrending. See issue 358 and issue 359.

Fixing a documentation bug (thanks to Matteo), small optimization in Sun and a few extra optimizations in MathOptim