CTF: pipeline integration

About

This PR aims at integrating CTF phase retrieval in the full-field pipeline.

To do

• Move generic code from ctf.py (ex. correct_spikes, recut)
• Correct docstrings (typos, etc)
• CtfFilter: add a method for doing the following steps: pad_interpolate, [norm by mean], CTF filter, "recut"
• Create a dedicated class involving flat distorsion correction (ex. for ID16a)
• Integrate in process config
• Integrate in FullRadiosPipeline
• Integrate in FullRadiosReconstructor
• End-to-end reconstruction test

Notes

The "ID16a CTF" consists in the following steps:

• dark subtraction
• estimate flat distorsion (flat - dark !)
• apply distorsion (interpn)
• median filter ("spikes correction")
• pad_interpolate
• retrieve phase
• "recut"

The "plain CTF" consists in the following steps:

• normal flat-field normalization
• spikes correction
• pad_interpolate
• normalization by mean (padded_im /= padded_im.mean())
• retrieve phase
• "recut"