Gain model specification, remove 1T defaults

Created by: JelleAalbers

This:

• Adds the gain_model and hev_gain_model options to describe PMT gain models (and replace the old to_pe_file option). These must be set to a tuple: (a model type string, a configuration or version specific to that model). The software high-energy veto has a separate option, because we may want to set this during online processing and never change it. Currently we only support a few basic model types; in the future time- or voltage-dependent models could be added:
  • to_pe_constant: configuration is a single number, the ADC -> PE conversion factor to use for all channels.
  • to_pe_file: configuration is a URL (or filename) pointing to a .npy file mapping run_id's to to_pe arrays.
  • disabled: configuration should be None. This is a placeholder for hev_gain_model to denote that the software-HEV should be disabled. (You can also disable the software-HEV by setting its threshold to a False value, but then it would be odd to still have to specify a gain model.)
• Replace the XENON1T-like defaults in PulseProcessing by values more sensible for XENONnT first data (no pulse filter, no high-energy veto, generous hit extension, uniform hitfinder threshold and gain). The XENON1T settings are moved to straxen.common.x1t_common_config, which is auto-applied for 1T contexts. I hope we can move all 1T-specific defaults here eventually so it doesn't clutter straxen later on.
• Removes straxen.n_tpc_pmts. The number of PMTs has for some time now been specified as part of the config.
• Lets mini-analyses takes config as an argument. They could already get it through context.config, but this seems more convenient.

These changes are inspired by Joran's #69.