issues with gprofiler2/gost

Have you checked the docs?

• nf-core website: troubleshooting
• nf-core modules documentation

Description of the bug

Several issues found for module gprofiler2/gost:

1. Inconcistency between using --organism mmusculus or not
   Running with --organism mmusculus vs running with --organism null --gmt_file gprofiler_full_mmusculus.ENSG.gmt (gmt obtained by downloading from their website) will return different results.
2. Fixing the database using set_base_url and then run gost on an archived database works locally but somehow gives error in the CI (https://github.com/nf-core/modules/actions/runs/13199316156/job/36850361811)
3. By default, the module will not use the gmt_file given by user, if organism is specified. Maybe it should be the other way around?

Other issues that makes testing the module using nf-test difficult:
4. Fixing the database using set_base_url and then run gost on an archived database is extremely slow even for test dataset (about 8 min vs 30s when using the updated database). Not sure why, but it could be related to how gost interact with old archives.

Start with the gprofiler2/gost module and the failing CI job 13199316156, then reproduce the nf-test cases using organism, gmt_file, and set_base_url. Compare the results and runtime for current and archived databases. Done means the input precedence is defined, results are consistent, archived-database tests pass in CI, and the slow case is understood or addressed.