spark_bids: dataset has to be a subdirectory of the cwd

[glatard]

Otherwise it doesn't get mounted to the container.

[glatard]

1. Extract the directory where the dataset is located. For instance, extract /a/b from /a/b/dataset.
2. Pass this directory to bosh using option -v, e.g., -v /a/b:/a/b so that it gets mounted in the container.