Fix digitizer saturation check

[zhut19]

What is the problem / what does the code in this PR do
Fix a bug that's causing pulse saturation check not working. The right condition should be max_in_record >= int(baseline), as record = int(baseline) - raw_record and pulse saturation results into raw_record = 0

[JoranAngevaare]

Looking at the code and the dtypes, I think this makes sense, thanks Tianyu.

Do you want me too have a look in the data too or have you found/done sufficient evidence and tests such that you are confident?

Other than that, I think we might want to increment the Peaklets version for this right?

[zhut19]

Thank @jorana, you are welcome to check, so there's less room for a mistake again. A few more things to do for this to work again (I mean before we bump the Peaklet version).

• Fixing another baseline issue in straxen _peak_saturation_correction_inner see Fixing saturation correction bugs XENONnT/straxen#541
• Allow partial computation of peak width in strax compute_widths see Add select index to compute width #465
• Add width computation after saturation correction in Peaklets see Compute width again after saturation correction XENONnT/straxen#542