Wrong image metadata (gdalinfo) in GetCoverage responses

[schindlerf]

Some GetCoverage test cases produce images which have wrong image metadata when queried with gdalinfo.

E.g:

< Files: expected/WCS20GetCoverageRangeSubsetIndicesDatasetTestCase.tif 
--- 
> Files: responses/WCS20GetCoverageRangeSubsetIndicesDatasetTestCase.tif 
13,14c13,14 
< Origin = (8.487755775500002,46.249103546100010) 
< Pixel Size = (0.031332727245640,-0.031300037595089) 
--- 
> Origin = (8.487755799999999,46.249103500000004) 
> Pixel Size = (0.031390966419295,-0.031370060044643) 
24,27c24,27 
< Lower Left  (   8.4877558,  32.2266867) (  8d29'15.92"E, 32d13'36.07"N) 
< Upper Right (  25.3760958,  46.2491035) ( 25d22'33.94"E, 46d14'56.77"N) 
< Lower Right (  25.3760958,  32.2266867) ( 25d22'33.94"E, 32d13'36.07"N) 
< Center      (  16.9319258,  39.2378951) ( 16d55'54.93"E, 39d14'16.42"N) 
--- 
> Lower Left  (   8.4877558,  32.1953166) (  8d29'15.92"E, 32d11'43.14"N) 
> Upper Right (  25.4074867,  46.2491035) ( 25d24'26.95"E, 46d14'56.77"N) 
> Lower Right (  25.4074867,  32.1953166) ( 25d24'26.95"E, 32d11'43.14"N) 
> Center      (  16.9476212,  39.2222101) ( 16d56'51.44"E, 39d13'19.96"N) 

This problem might be related to Ticket #37.

[krauses]

There are two problems related to this ticket

1. In the fixtures, the extent has been rounded to 10 significant digits (i.e. 7 digits behind the comma). This was not the case in previous versions, so the difference in the origin coordinates is due to these rounding errors.
2. Apparently, there has been some miscalculation of image sizes before. In other words the expected outputs have been wrong for quite some time.