Make matching of fieldmaps even more strict?

[yarikoptic]

We are chatting with @raamana and @sinhaharsh about fieldmaps and their "horizontal checking" in https://github.com/Open-Minds-Lab/mrQA/ . mrQA ran on one of our converted with heudiconv datasets caught that some fieldmaps were not entirely kosher. mrQA gave such a report:

EchoTrainLength	82,	83,	0028_01,
PhaseEncodingSteps	82,	83,	0028_01,
SequenceVariant	SK,	SK\OSP,	0028_01,

so it considered all those 3 fields. And indeed EchoTrainLength (and the two others) seemed to be a little off in one case whenever Shim settings were the same (according to my eye):

[bids@rolando 1076_spacetop] > grep EchoTrainLength sub-0028/ses-01/fmap/sub-0028_ses-01_acq-mb8_*_run-01_epi.json
sub-0028/ses-01/fmap/sub-0028_ses-01_acq-mb8_dir-ap_run-01_epi.json:  "EchoTrainLength": 83,
sub-0028/ses-01/fmap/sub-0028_ses-01_acq-mb8_dir-ap_run-01_epi.json:      "EchoTrainLength": 83,
sub-0028/ses-01/fmap/sub-0028_ses-01_acq-mb8_dir-pa_run-01_epi.json:  "EchoTrainLength": 82,
sub-0028/ses-01/fmap/sub-0028_ses-01_acq-mb8_dir-pa_run-01_epi.json:      "EchoTrainLength": 82,

[bids@rolando 1076_spacetop] > grep Shim sub-0028/ses-01/fmap/sub-0028_ses-01_acq-mb8_*_run-01_epi.json
sub-0028/ses-01/fmap/sub-0028_ses-01_acq-mb8_dir-ap_run-01_epi.json:  "ShimSetting": [4171, -7765, -7131, 742, -32, -105, -201, -3],
sub-0028/ses-01/fmap/sub-0028_ses-01_acq-mb8_dir-pa_run-01_epi.json:  "ShimSetting": [4171, -7765, -7131, 742, -32, -105, -201, -3],

In #482 (by tireless @pvelasco ) we agreed to rely on Shim and geometry, but we might consider those few others to be added for the most stringent criteria?