adding support 3/4 sampling for 40MHz (in that case OFDM symbols are not necessarily aligned on 32byte -> segfault on AVX2 architectures)

* To run gNB : sudo ./nr-softmodem -E -O ../../../targets/PROJECTS/GENERIC-LTE-EPC/CONF/gnb.band78.tm1.106PRB.usrpb210.conf
* To run UE : sudo ./nr-uesoftmodem -E --numerology 1 -r 106 --phy-test -C 3510000000 --ue-rxgain 90 -d
* We had done a wired test with 40dB attenuator in between gNB and nrUE.

Nr ue fapi improvements

See merge request oai/openairinterface5g!618

however, I had to hardcode the parameters for the PUSCH in nr_ulsim as the fapi interface still has bugs.

This reverts commit 3574481b.

Signed-off-by: Raphael Defosseux <raphael.defosseux@eurecom.fr>

Signed-off-by: Raphael Defosseux <raphael.defosseux@eurecom.fr>

Signed-off-by: Raphael Defosseux <raphael.defosseux@eurecom.fr>

Integration develop nr 2019w24

See merge request oai/openairinterface5g!615

Merge remote-tracking branch 'origin/rfsimulator-simulate-channel' into integration_develop-nr_2019w24

Conflicts:
	openair1/SCHED_NR/phy_procedures_nr_gNB.c
	targets/RT/USER/nr-ru.c

 - dci_request and dl_indication for PBCH now also on stack
 - fixed type of nrUE_config (was using structure for gNB before)
 - added configuration of PBCH to phy_config_request and using this config now in phy_procedures_nr_ue
 - we need to add configuration of PDCCH and PDSCH also to phy_config_request (there are still some hardcoded parameters)
 - removed UL scheduling for now

- This was due to alignment problem in rxdataF_ext when odd number of RBs were assigned