Download HCP rest + task data from the S3 repo, and runs the nistats/FSL GLM on time series.

python setup.py install
cd examples
python download_tasks.py
python download_rest.py

HCP_data: root of the HCP dataset, where it will be downloaded
HCP_AWS_KEY: for S3 bucket access
HCP_AWS_SECRET_KEY: for S3 bucket access
CDB_USERNAME: for behavioral access
CDB_PASSWORD: for behavioral_access

• write tests
• automate restart when failures
• download non-completed subjects
• write functions to copy resources in the HCP_DATA dir
• move nilearn like fetchers from modl into the package. For now, you may use development branch of github.com/arthurmensch/modl
• merge downloaders and fetchers as in nilearn (?)