cmake: Fix CMP0175 warning by adding POST_BUILD to TARGET form of add_custom_command()

[echoix]

I tried to explore using a cmake build for generating documentation, but fails for addons. But when exploring, I found a simple fix to one of the warnings that are displayed since CMake 3.31. Starting from CMake 3.31, more rigourous argument checking is done, and it complains that with the TARGET form of add_custom_command(), the type of event must be specified, like POST_BUILD.

This PR adds POST_BUILD, the default assumed when it was missing.

See https://cmake.org/cmake/help/latest/policy/CMP0175.html

[echoix]

I thought the windows failure was only a fluke, as I saw on main and on my branch, even after reruns. So, it doesn't look like a transient thing.

[echoix]

Does someone have any idea on the windows failures since 3 days? It's completely blocking us, unless we lower our thresholds.

On multiple occasions, a passing commit before the constant failures, once rerun (more than once, on a couple days span) fail the same way. => it is not some code on our part.

I checked really hard, and couldn't find a cause. I diffed the CI logs of a passing job, and a failing rerun 4 days later.
I don't see any package differences in OSGeo4W (and no new commits in the repo, but he has the habit of pushing the changes days or weeks later :( ). No Python package changes no Python version changes.

The only difference is in the msys packages. The only one that is a minor version change is https://github.com/fmtlib/fmt that passed from 11.1 to 11.2. Other than that, it is mostly some recompiles that changed.

So, I really don't see what could be going on. Is it something like a DLL collision that needs OSGeo4W to recompile? It's weird, the error is in Python library code, mentioning that the file is empty. When doing a full read diffing the two logs (even with parallel output with different order), I didn't catch a different message in the compile log, that would point to an error when compiling.

[echoix]

I've attached to the runner by ssh (using tmate), before running the tests. The errors in gunittest seem to be legitimate, as the "--interface-description" call doesn't work for all binaries, but works for some.

Since I wasn't able to copy the text properly (it is using tmate), I had to take screenshots. I have more for my tries on enabling loader snaps (https://learn.microsoft.com/en-us/windows-hardware/drivers/debugger/show-loader-snaps, for example: gflags /i i.fft.exe +sls) and running with cdb.exe

r.tile works, i.fft doesn't (these were simple to type)

C:\OSGeo4W\opt>c:\OSGeo4W\opt\grass\grass85.bat --exec i.fft --interface-description                                    
WARNING: Concurrent mapset locking is not supported on Windows                                                                                                                                                                                  
C:\OSGeo4W\opt>c:\OSGeo4W\opt\grass\grass85.bat --exec r.surf.gauss --interface-description                             
WARNING: Concurrent mapset locking is not supported on Windows                                                                                                                                                                                  
C:\OSGeo4W\opt>c:\OSGeo4W\opt\grass\grass85.bat --exec i.pca --interface-description                                    
WARNING: Concurrent mapset locking is not supported on Windows                                                                                                                                                                                  
C:\OSGeo4W\opt>c:\OSGeo4W\opt\grass\grass85.bat --exec i.smap --interface-description                                   
WARNING: Concurrent mapset locking is not supported on Windows                                                                                                                                                                                  
C:\OSGeo4W\opt>c:\OSGeo4W\opt\grass\grass85.bat --exec v.lidar.correction --interface-description                       
WARNING: Concurrent mapset locking is not supported on Windows                                                                                                                                                                                  
C:\OSGeo4W\opt>c:\OSGeo4W\opt\grass\grass85.bat --exec r.gwflow --interface-description                                 
WARNING: Concurrent mapset locking is not supported on Windows                                                                                                                                                                                  
C:\OSGeo4W\opt>c:\OSGeo4W\opt\grass\grass85.bat --exec r.sim --interface-description                                    
WARNING: Concurrent mapset locking is not supported on Windows                                                          
'r.sim' is not recognized as an internal or external command,                                                           
operable program or batch file.                                                                                         
Press any key to continue . . .                                                                                                                                                                                                                 
C:\OSGeo4W\opt>c:\OSGeo4W\opt\grass\grass85.bat --exec r.sim.water --interface-description                              
WARNING: Concurrent mapset locking is not supported on Windows                                                          

C:\OSGeo4W\opt>   

Details