tof compensation issue in case of empty CXX columns

[philippebrocard]

prx/src/prx/main.py

Line 261 in 3667497

# satellite clock offset plus those small terms of a few tens of nanoseconds

I suspect this line is causing issues in some rinex obs files with empty CXX columns that seem not parsed as nan but something else (zeros ?). Then the tof becomes erroneous, leading to satellite coordinates inconsistent with the sat position at transmission time. Can be solved by using the maximum pseudo range (or any non 0 one).

[jtec]

Can reproduce the issue by setting a breakpoint here

prx/src/prx/main.py

Line 253 in 0101d7e

tof_dtrx = pd.to_timedelta(

and running

per_sat[per_sat[code_phase_columns].sub(per_sat[code_phase_columns].mean(axis=1, skipna=True), axis=0).abs().max(axis=1) > 1000]

which shows a bunch of satellites with wildly different code observations, mostly GLONASS and one Beidou: