Add the possibility to delete keys based on string match (from the begining or anywhere in the key). This is useful for example to remove all stations declared from the user in the ini file.

After talking with the GSN team, the GSN plugin has been fixed and works again. It has been simplified according to their recommendations (ie that the metadata is always delivered with the data). We should still implement arbitrary units handling (parsing the strings and converting to SI).

In order to make it more modular and allow more advanced processing (virtual stations, downscaling, multiple io sources) the IOManager is now a proper wrapper: the processing is fully delegated to either a GridsManager or a TimeSeriesManager. These can in turn be reused at some other places (like in the Meteo2DInterpolator) to offer full buffering and processing of "local" time series. The buffering is now done by these two new classes (therefore the BufferedIOHandler class has been removed) and it is planned that dataClasses will be implemented for proper ringbuffers (one for time series, one for grids and one for isolated points) and receive the generic buffering code later on. 

A new fastPow(double, int) has been implemented for positive exponents. The begining of a generic handling of units also appeared in IOUtils. Finally, the ProcessingLevels are now defined in IOUtils instead of IOManager.

Now the Winstral algorithm properly handles the case of only one station providing HNW with the default base algorithm (by switching to CST in this case). The availability of wind data is now part of the rating return by Winstral. The case of no precipitation (for Winstral) as well as no wind (for Liston and Ryan) are now also handled in the rating method.

When a plugin was throwing an exception, the mapPlugins could get corrupted in the (rare) case of the caller catching and moving forward (uncaught NULL pointer). This is now not anymore possible, by design.

The simpleWindInterpolate spatial interpolation was not properly handling nodata values. Therefore the 2D_interpolations reference file for VW had to be corrected. The statistical tests now work better with nodata values (setting values to IOUtils::nodata instead of hard-coding -999 in the test data sets).

Fixed a bug in a (currently) unsued optmized method for nth_root of a number (negative numbers were not working)

When using Snowpack simulations in the database for computing precipitation, this was leading to lots of very small events. This should now produce less events but with more precipitation.

The nodata points are now always properly filtered out when computing a least square fit. The code has been a little reshaped in order to be closer to the reference document that has also been cited. Some comments have been added. The quadratic fit was wrong and has been fixed.

The statistical tests now cover the least square fits and ad-hoc linear regressions.

The ARC plugin properly builds the file names for reading and writing grids with the "A3D_VIEW" option. 

When introducing TAU_CLD, a nasty bug found its way into the ALLSKY_LW generator that lead to always compute a perfectly clear sky... 

A new Date::getTime() method has been written. A better error message has been implemented for ProcHNWDistribute. The daily_solar temporal interpolation can now handle daily sums provided at midnight (then assuming they belong to the day that just finished) and the documentation has been updated.

The grids.isEmpty() method has been renamed as empty() in order to follow the STL convention. The ARC grids written with Alpine3D naming were still not properly named, they should now be directly recognized by the Alpine3D grid viewer.

The getTanMaxSlope method used for Winstral or Ryan has been optimized and improved: now a pixel that would be just behind a ridge would be properly processed (ie it would see the uphill slope and not the downhill slope behind the ridge).

Adding the cmake script to detect MySQL. This is currently only working on Linux (no proper paths for Windows).

Small complement to the "filter dev" documentation. A first implementation of the wet globe temperature has been contributed, but this is NOT usable yet (a few coefficients still must be set for the convective heat transfer coefficient).

PINT is now properly supported in SMET. The OLWR/TSS conversion must now be handled by a data generator (implemented as "TS_OLWR"). A bug in the Coords documentation has been fixed.

The splitting was computed using TOA_beam when it should have used TOA_h (after checking the Reindl, 1990 paper)

The GSN_URL key is now mandatory and the GSN_USER and GSN_PASS are now optional (for publicly accessible sensors)

A bug has been found (thanks to ctest!) that was introduced when we added TAU_CLD: the new MeteoGrid SHADE was only partially implemented and this lead to segfaults. The decrementing code in ARCIO was wrong and this has been fixed.

Fixing two tests that were not working anymore (since a new parameter, TAU_CLD had been inserted). White spaces in MeteoData.h, more logical iteration in ARCIO.

The basic operators have been implemented in Grid2DObject and Grid3DObject as well as DEMObject (with the proper update strategy). 

In order to suppress the risk of out of date "ncols, nrows" and since these were redundant with Array2D<>.nx/ny, they have been suppressed and it is therefore now mandatory to rely on the getNx/getNy/getNz getters. This is much safer but impacted quite a lot of code... 

Adding a new meteo parameter: the cloud transmissivity TAU_CLD defined as iswr_meas / iswr_clear_sky (ie the potential radiation computed at ground level with a standard atmosphere, see Iqbal 1983). This parameter is already used by the AllSky_LW and PotRad generators and properly handled by the PNG plugin. The PNG plugin can now also properly output a hillshade (as MeteoGrids::SHADE).

A bug has been fixed in the Winstral implementation that could lead to infinite loops for very small DEMs.