Added option for bipolar re-referencing before analysis

[usuario1marc]

Added

• New option in the preprocessing configuration: bipolar_montage, a boolean which defaults to False. When True, input channels will be re-referenced to a bipolar setting before computing metrics.
• The process_subset() function from _erdetect.py now checks this option, and if True, uses load_data_epochs() instead of load_data_epochs_averages(), and applies a bipolar rereferencing.
• Expanded the information in the output_dict to also return the stimpair_indices and the raw trials if the montage is bipolar.
• New function in utils/misc.py: mount_bipolar, which takes as an input monopolar channels and turns them into a bipolar setting.

Modified

• Output directory is subdivided in session and run.
• Changed version to 2.7.0

Hello! This is my first time doing a PR, so excuse me if I used non-orthodox language for explaining my changes. I used ERDETECT in an internship for my Biomedical Engineering bachelor in a project about locating CCEPs. I had to modify the module to be able to use it in a bipolar setting. After reaching out to Dora Hermes to talk to her about this project, she encouraged me to make a pull request, so here I am. I hope you can find it useful. Cheers!