Lazy loading CompartmentReport before subprocess hangs

## Issue
I have an issue on BB5 where I need to use `brion` in combination with `py-bluepy` modules to read some reports and then execute other process with **subprocess**.
The issue is that after using **brain** to get some information, then **subprocess** hangs.
We were able to identify that if we load `from brain import CompartmentReport` at the beginning of the `test_brion_report.py` file (so it is not being lazy loading) the issue disappears.
But BluePy uses this with lazy load and anyhow it should work the same so, can you investigate why am I getting this issue?

I past below the files examples and the script to run this test.

## Save files
**run.sh**
```
#!/bin/bash
module load brion
module load py-bluepy
python ./test_brion_report.py
```

**test_brion_report.py**
```
# from brain import CompartmentReport

t_start = 0
t_end = 300
t_step = 0.1
report_path = '/gpfs/bbp.cscs.ch/home/antonel/unicore_jobs/47f40ebc-8910-4fb5-99ad-d5fcc28ff059/5Cells_report_0.bbp'

def use_brain():
import pandas as pd
import numpy as np
import subprocess
from brain import CompartmentReport

# obtained from the original brion code
def _wrap_index(index):
""" Create pandas (Multi)Iindex from 'raw' index. """
return pd.MultiIndex.from_tuples(index)

report = CompartmentReport(report_path)
view = report.create_view([51848])

print('Calling subprocess ---------------------')
subprocess.call(['ls'])
print('Done')

use_brain()
```
## Allocate resources on BB5
`salloc -n 1 --time 2:00:00 --account proj42 -C uc3 --partition prod`

## Run script
`sh run.sh`

Start with run.sh and test_brion_report.py, then reproduce the difference between the commented top-level CompartmentReport import and the import inside use_brain(). Run the script with subprocess.call(['ls']) and confirm that the hang is resolved without eager loading.

Written by the indexing model from the issue text.