Tables locking

[OlivierHecart]

No description provided.

[Mallets]

The proposed changes allow to reduce the time a write lock is held on the tables upon reception of new declarations. However, the proposed changes seem to introduce a performance regression on the machine used to mesure Zenoh's performance (see https://zenoh.io/blog/2023-03-21-zenoh-vs-mqtt-kafka-dds/).

From 4.35M msg/s -> 4.05M msg/s with a data payload of 8 bytes (i.e., 300K msg/s went missing). I believe this should be further investigated before merging.

[Mallets]

I've been investigating and rerunning the tests, it seems a fresh compilation provides better results. I still experience a small drop about 20K msg/s but I believe they are negligible compared against the benefits brought by the new locking strategy.