Restore subject 32 46 and 49 of Cho2017 dataset

[carraraig]

Reintroduction of excluded subject 32 46 and 49 from Cho2017 dataset

[sylvchev]

Thanks @carraraig
Note for future self and future maintainers: we investigated the reason why those subjects were removed. Git blame indicates that Alex Barachant made this choice but it was not documented. I visually inspected the data and made some basic evaluation with a pipeline, no error. I dig in the reference article for this dataset, they mentioned that 2 subjects have high correlation with EMG but the identifier are s29 and s34 (paper), not s32, s46 and s49 (code). Here is an excerpt:

Two subjects (s29 and s34) showed that more than 90% of their trials were correlated with EMG; most of the trials demonstrated a greater power of high-frequency EMG (50–250 Hz) in the task-related period after onset than the resting period before onset. Thus, these two subjects were declared bad subjects and were discarded in the further analysis.