PRB based averaging can activated from command line options

A second thread message is created for L1 tx processing and pushed into the threaPool.
The message contain the nfapi rel15 PDUs for SSB, PDCCH and PDSCH. The memory of these PDUs are reused for subsequent slots.
Need to fix the nfapi mode VNF PNF implementation to adapt to this memory scheme.

changes in gtpu interface for testing, small changes in L1 for PUSCH signal/noise measurements. SRB2 interfaces.

for gNB its in L1 section of config file
for UE its a cmd line paramenter

added n1L1_stats.log. improvements to noise level (PUCCH/PUSCH) and signal level (PUCCH) computations. added DTX detection parameters for PUCCH/PUSCH/PRACH in L1 section of configuration file. hotfix from C. Roux for SRB PDCP integrity. extra logging to debug missing PUSCH/DCI at UE.

lowmem option for allocating transport channel buffers in NR_DLSCH/NR_ULSCH. Used for unitary simulation in CI.

Leaving ru_thread and L1 stats thread

Created 3 threadpools for L1 rx, L1 tx and RU tx processing

- set the proper platform variables definitions
- cleaned up commented out code

The LDPC lifting size has to be stored for each harq_process as its only assigned in the first round.

The BG was set to its default for rounds > 0. This is now fixed by adding it to harq structure.
Also a fix for wrong I_lbrm set in dlsch decoding.

- Most DL code is reused for Ul also to reduce code duplication
- Common CPE calculation function is used for error estimation
- Common slot processing function is used for inter/extrapolation
- Removed old implementation dedicated to UL

adding beamforming support to FAPI: MAC fills ssb_beam_id_list in ssb config. PHY uses this info to fill new beam_id field. This is further copied to RU and applied there. Nothing is tested so far.

This commit contains work done by Eurecom and by Benetel.

- New Data Structure to store PTRS RE's is removed
- LLR's are calculated at the end symbol for whole slot