Options to pass as variable:
    "rms", "trace" and "spec"

Options to pass calibration:
    "linear" and "full"

Notes to self:
use as event selection:
    type: "PHYSICS"
    trigger rate < 2 Hz (higher trigger rates are probably windy/human activity periods)

BROKEN RUNS:
station 24, run 363, event 105 is broken (200 header entries, but only 100 waveforms)

Calibration:
Currently the full calibration only works in the uproot backend, pyroot struggles with memory issues
Hence the code only uses the uproot backend for now (we also don't have VolCal files)

Potential artifacts:
    Have a look at ringing, which can be caused be the deconvolution of the detector out of the trace

What to do with G = 0? this represents e.g. a bandpass filter.
This information is lost so what value should be used as placeholder or this should be deleted?
For the purpose of thermal noise from ice I assume these can be set to 0 since the signal here is purely electronics noise?

BE CAREFUL WHEN TAKING MEANS
When looking into noise taking the mean over many noise events will yield a trace very close to 0. This because the random fluctuations interfere destructively. One should only take the mean over the final calculated variable.

USE OF CODE
Do not forget to add the root directory "noise_study" to the pythonpath e.g. in an alias