Add support for custom_rosdep_urls in build files

[j-rivero]

In addition to the existing support of the parameter custom_rosdep_urls for the doc and source files, the PR adds the support for the build files so it can be used for things like nightly jobs.

I tested it here https://citest.build.osrfoundation.org/view/Rci/job/Rci__nightly-release_ubuntu_noble_amd64/26/ which builds gz-rendering8 (outside of ROS) using the packages.osrfoundation.org repository and an external rosdep rule to resolve libogre-next-2.3-dev to the package of the same name hosted in packages.o.o.

[cottsay]

Thanks for adding documentation.

The block you added the documentation to is specific to CI jobs, but the file you added the actual configuration support to is global to many job types. I think it might be more appropriate to add the config to ci_build_file.py rather than the global build_file.py.

Alternatively, we could keep the config change in build_file.py and move the documentation block, but in that case I think we would need to remove the configuration and documentation in the other specific jobs that already have it, and now inherit support from build_file.py. That said, I'm not sure that every job type that inherits from build_file.py can use rosdep to begin with, so I'm a little wary of this approach.

[j-rivero]

The block you added the documentation to is specific to CI jobs, but the file you added the actual configuration support to is global to many job types. I think it might be more appropriate to add the config to ci_build_file.py rather than the global build_file.py.

Alternatively, we could keep the config change in build_file.py and move the documentation block, but in that case I think we would need to remove the configuration and documentation in the other specific jobs that already have it, and now inherit support from build_file.py. That said, I'm not sure that every job type that inherits from build_file.py can use rosdep to begin with, so I'm a little wary of this approach.

Thanks for the hint and the context. I agree on going conservative and restrict the change to the ci_build_file instead of trying to push a more general approach that we are not sure how it would go in some scenarios.

Changed the code in 7884b1a

[cottsay]

Looks good, thanks!