✨ [rtl] add simple branch prediction

[stnolting]

This PR adds something like a very branch prediction (predict "always taken") to the CPU front-end (fetch engine).

Whenever a branch or jump instruction is being executed by the execute engine, the fetch engine will stop fetching further instruction data until the branch target address has been computed (also in case of a non-taken conditional branch).

• reduces branch penalty: no need to wait until a "speculative" instruction fetch has been completed since the fetch engine waits for the branch/jump instruction to retire
• reduces bus traffic (eventually blocking data memory access)
• higher performance