Correct SA3 slow data by SASE1 contribution

• correct SA3 slow data by the SASE1 contribution based on the information from Theo (https://in.xfel.eu/elog/SCS+Beamline/3640)
• adds the slowTrain lowpass averaged data from the XGM ADC
• renames the mnemonics to something more systematic in terms of the use of SA3 to identify the bunches and not the XGM device

Unfortunately, the numbers don't check out right now with what is shown in the FEL Status Display: https://in.xfel.eu/elog/SCS+Beamline/3643

Email sent to Theo and Laurent about it. Once the problems are solved we can look into merging.

It seems the problem was in the ADC saturating. (https://in.xfel.eu/elog/SCS+Beamline/3669)