threading: Add a pending list for async task insertion

This slightly reduces the contention of the main lock by adding subsequent tasks to a separate pending list.

Helps with higher thread count, basically allowing for better cpu usage, improving fps for these configurations.

benches: https://drive.google.com/drive/folders/1OWqZPC_CJCeY9WmVpW2JYZJrsn272Dzo?usp=sharing