Fix the inconsistent behaviour of --participant_label

[t-sanchez]

Selecting multiple participant labels at once has been reported to make the resulting reconstruction fail.

Investigating to try to reproduce the bug.

[t-sanchez]

For example : when running subjects 5 and 12, the output has files named sub-05 in the sub-12 directory.

[pdedumast]

@t-sanchez can you try again from branch v2.1.0? Might have been solved with nipype upgrade... 🤞

[t-sanchez]

Sadly, this still seems to occur sometimes.

In this run, the desc-motion_stats.tsv of chuv018 was fond in all other folders of the run. It's also the case for the corresponding figure.