Release 0.4.0 (#110)

* Fix building tests in multi-gpu environment (#98) 
* Change cmake.verbose = true to build.verbose = true (#99) 
* Use build-system.requires to set scikit-build-core minimum version
(#100)
* Set CUDA version in one file (and use everywhere else) (#101) 
* Drop Python 3.9 support (#102) 
* Use CI workflow branch 'branch-24.10' again (#105) 
* Use conda strict channel priority. (#109) 
* Update to CUDA 12.6 (#97)

pynvjitlink/VERSION

@@ -1 +1 @@
-0.3.0
+0.4.0