Add documentation on PRACH processing

doc/rach_processing_in_gNB.md

+The gNB RACH processing
+
+1. Collect I/Q samples
+The gNB/ru main loop, in ru_thread()
+If there is rach occasions in this slot
+this main loop calls rx_nr_prach_ru() that stores the I/Q samples in the prach_item_t context
+
+2.process the samples
+L1_rx_thread() has received slot indication by msg on queue resp_L1, it calls rx_func() => L1_nr_prach_procedures()  
+L1_nr_prach_procedures() fills nfapi_nr_rach_indication_t structure from result of rx_nr_prach() that process the content of rxsigF[prachOccasion][aa]
+rx_func() sends the decoded rach to upper processing by call to NR_UL_indication() that calls handle_nr_rach() => nr_initiate_ra_proc()  
+
+nr_initiate_ra_proc() creates a UE context if it needs, then sets UE.ra.ra_state to what it should do
+
+3. send the RACH DL
+tx_func() that is launched by L1_tx_thread() as it gets L1_tx_out msg (from main loop in 1.)
+tx_func()=>run_scheduler_monolithic()=>gNB_dlsch_ulsch_scheduler()=>nr_scheduler_RA()
+nr_scheduler_RA() process all existing UE contexts, to see it it has to schedule a RACH DL
+
+4. global synchronisation
+resp_L1 msg for (2.) is created in tx_func() that is launched by L1_tx_thread() (see3.) when it receives slot indication as it gets L1_tx_out msg
+this msg is sent by the main loop (see 1.)
+tx_func()=>run_scheduler_monolithic()=>nr_schedule_response ()=>nr_schedule_ul_tti_req ()=> nr_schedule_rx_prach()
+writes in the triggering structure to (1.) that it will have to run  rx_nr_prach_ru() in a given future slot