for 8 port version generate first 4 GTXs with BUFR to reduce delay asymmetry between 8 port and 18 port version

RTU: 1) not forwarding to CPU urecognized broadcast; 2) not going through LEARN_SRC when DST_MAC not found; 3) correcting pass_pbdu behavior (overrides pass_all only if pass_all is FALSE)

RTU: don't send ureq (learning interrupt and datat for CPU) when destination MAC is not found in HTAB - it seems to have no sense

RTU-bugfix: aging_bug-related two changes 1) the source entry search in HTAB is considered successful only if the found port_dst_mask matches the search-requesting port; 2) updating aging table only on source MAC found. No changes for destination MAC lookup

top levels: added I2C master for TMP101 onboard temperature sensors and rewired pll_status_i to v3.2. dedicated clock input

rt: synchronize slave channel DDMTD counter before starting MPLL (fixes reversed DMTD locking issues)