The Processing Threads limit is the sum of threads available for federation jobs plus the short-lived HTTP requests made to your Mastodon server.
Example: If your plan has a limit of 4 Processing Threads. Your server can only perform 2 concurrent federation jobs and 2 concurrent short-lived HTTP requests simultaneously.
When all the Processing Threads are occupied, new requests will be queued.
Using the previous example, where a limit of 2 concurrent short-lived HTTP requests is in place if 3 people attempt to request a page load the web interface at the same microsecond, 2 will immediately be processed. The third will need to wait (in this case, it should take microseconds).