RegTests with epoll failrues

[kostasrim]

See https://github.com/dragonflydb/dragonfly/actions/runs/12764061257/job/35575450779#step:6:3437

• test_migration_timeout_on_sync
• test_network_disconnect_during_migration
• test_cluster_migration_huge_container
• test_reply_count
• test_big_containers
• test_cluster_migration_while_seeding
• test_denyoom_commands

For test_big_containers it would almost never finish because of a bug within the epoll socket (romange/helio#363). After the fix, locally the memory does not surpass 1.4x (dropped from 2x) but it seems that on the CI it still fails with (2x). I also:

• compared the INFO MEMORY from both EPOLL and IOURING leading to very similar memory footprints for both variants
• calculated the total allocations and deallocations with the memory tracker for both EPOLL and IOURING and both of them had very similar results

I suspect there is something around the EPOLL socket operations that "wastes" memory but we need to somehow narrow it down

[kostasrim]

Full list of skipped tests: https://github.com/dragonflydb/dragonfly/pull/4426/files

[kostasrim]

Two more: test_cluster_migration_while_seeding https://github.com/dragonflydb/dragonfly/actions/runs/12875654869/job/35897346679 and https://github.com/dragonflydb/dragonfly/actions/runs/12840327559/job/35808838383

[kostasrim]

https://github.com/dragonflydb/dragonfly/actions/runs/12873612142/job/35891503547 test_denyoom_commands