0% coverage uploads (and wasted time) coming from archive_dist_test_debug:

https://spruce.mongodb.com/task/mongodb_mongo_master_AL2023_arm64_coverage_archive_dist_test_debug_patch_502f742565beacfd2fa85386f6352b2b88bec242_66ffd662dbf0f30007ac47fe_24_10_04_11_50_16/logs?execution=0

https://parsley.mongodb.com/evergreen/mongodb_mongo_master_AL2023_arm64_coverage_archive_dist_test_debug_patch_502f742565beacfd2fa85386f6352b2b88bec242_66ffd662dbf0f30007ac47fe_24_10_04_11_50_16/0/task?bookmarks=0,19768

We are sniffing for gcno/gcda sort of files to see if we should collect and upload coverage. Maybe we need something better to look for.

The script is run in tasks outside of the codecoverage build variants, but there is definitely no .gcno data there, so the early return on those should still be well-behaved. It's just the misc tasks in the coverage variants that could short circuit if we optimize.