kp.calib.Calibration incorrectly adds t0 to McHits time

Environment

• KM3Pipe version (km3pipe --version): '9.0.0b5'
• Python version (python --version): 3.6.8

Additional information

When using kp.calib.Calibration to calibrate McHits, t0 gets added even though it should not.

I converted a mupage root file using h5extract (this is not relevant for this issue). Then, in this dir /sps/km3net/users/sreck/storage/four_line/prod1/mupage/v5.40_km3/temp, I build a pipeline to calibrate the blob and print hits.time and mchits.time:

import km3pipe as kp
import km3modules as km


h5_file = "mcv5.40.mupage_10G.km3_AAv1.jterbr00005782.jorcarec.aanet.837.root.h5"
detx_file = "KM3NeT_00000044_00005782.v5.4_PMTeff.K40_muonCalibration_shift00.detx"

pump = kp.io.HDF5Pump(filename=h5_file)
calib = kp.calib.Calibration(filename=detx_file)
mc_time_cor = km.mc.MCTimeCorrector()

blob = next(pump)
blob = mc_time_cor.process(blob)
calib_hits = calib.apply(blob["Hits"])
calib_mchits = calib.apply(blob["McHits"])

mean_time = calib_hits.time.mean()
print("Hits time:\n", (calib_hits.time - mean_time)[:10])
print("McHits time:\n", (calib_mchits.time - mean_time)[:10])
print("McHits time - t0:\n", (calib_mchits.time - calib_mchits.t0 - mean_time)[:10])

Which gives this output:

Hits time:
 [-318.7426111   194.7713889   264.13038892   42.2343889   363.69238892
 -116.55261108   14.88638891  -71.0846111  -117.23061109 -114.21161109]
McHits time:
 [207810.02096483 207819.43096727 207839.37696606 207837.84097399
 207864.469974   207899.76296544 207900.88297582 207898.76397461
 207941.65597826 207943.20996971]
McHits time - t0:
 [-117.25603516 -156.56603272 -191.53603394 -191.60602601 -220.35602601
 -237.39603455 -236.27602418 -235.7460254  -248.83602174 -247.62603028]

So mchit times as they come out of calib.apply are wrong, as they are nowhere close to the real hit times. If I subtract t0 again, they are right.

--> t0 should not be added to mchits when calib.apply is run.

I also wonder if the functionality of km.mc.MCTimeCorrector should be a part of kp.calib.Calibration, since the times of McHits and Hits are not comparable without it (i.e. one could see it as a part of the calibration).