Fifo readout simplification and maintanance

This MR reverts to a more simple fifo readout from the past; deleting some unnecessary complications from multi-module "parallel processes" readout. It:

• Deletes the per receiver channel thread
• Removes multiprocessing to filter the raw data
• Add a data cache for fifo readout
• Fix and re-activate memory leak testing
• Remove some warnings
• Update docker runner containers
• Activate automatic scan testing for RD53B and fix 1-2 scans to work with RD53B
• Add numba cache for analysis (saves > 6s for each interpreter start)
• Fix #478 (closed) and #485 (closed)

If it turns out to be too slow for a quad module that is read out in parallel multi processing should be put back.