[ENH] Handle IQM distributions

[nikhil153]

Is there an existing issue for this?

• I have searched the existing issues

New feature

• Display group-level IQM distributions from the local dataset
• Display scanner-specific IQM distributions from the MRIQC API endpoint

Unclear documentation

No response

[laurapiro17]

between the two options here i'd lean towards pre-downloaded reference distributions (ABIDE, OpenNeuro) as default and the MRIQC API as optional. runtime network dependency breaks offline workflows and in a lot of clinical environments internet access is restricted anyway. it also makes the percentile computation reproducible across runs. has anyone looked at what normative datasets mriqc-learn already uses? might be something reusable there