Refactor confounds and QC metrics

[tsalo]

Closes none.

Changes proposed in this pull request

• Combine ComputeCBFQC and ComputeCBFQCforGE into ComputeCBFQC.
  • The only differences, as far as I could tell, were (1) the tissue probability threshold for GE scans was 0.8 instead of 0.7 and (2) mean FD and RMSD were only calculated for non-GE data.
  • Move ComputeCBFQC from aslprep.interfaces.cbf_computation to new aslprep.interfaces.qc.
• Lightly refactor some of the functions in aslprep.utils.qc. There's still a lot to do there though.
• Combine init_cbfqc_compt_wf and init_cbfgeqc_compt_wf into aslprep.workflows.qc.init_compute_cbf_qc_wf.
  • The only differences were that the non-GE version uses the confounds matrix and RMSD file, which don't exist for GE data.
  • Add a parameter, is_ge, to handle the differences.