Add "smartiness" for MP2RAGE support

[yarikoptic]

I think we should do "post-treatment" to provide generic support across heuristics that

• if there is mp2rage in suffix or _acq- after heuristic but we have multiple sequences named identically and coming from the same sequence name (and some UID groupping, relevant also for Add PDT2 handling #834 ), and we have series_description ending with variable suffixes:

then we could add some ad-hoc handling to fit where in BIDS for quantative anatomicals it should fall

ref : https://bids-specification.readthedocs.io/en/stable/appendices/qmri.html#unit1-images

use case of @jbpoline at BIC

[bpinsard]

If I am not mistaken (never got data from that WIP which seems to be compressed-sensing), this is not exactly the same case as for PDT2, as this results by default in separate series that can be dealt with at the heuristics level, provided the right level of custom metadata digging.