[RACL] Deal with CDC on policy distribution and error logging

In RACL, we asynchronously distribute the policy registers from `racl_ctrl` to the subscribing IPs. `racl_ctrl` and the subscribing IPs might be in different clock domains, causing possible CDC issues.

A similar CDC can arise when returning error information back from the subscribing IPs to `racl_ctrl`. We might have to sync the error trigger signal and then have an async FIFO on the error information.