Add a silent issue if dates in _scans are not anonymized

[chrisgorgo]

Check if the dates are from 1900 or later. If so raise the issue. Explain anonymization via random time shifts

From BIDS spec:

Dates can be shifted by a random number of days for privacy protection reasons. To distinguish real dates from shifted dates always use year 1900 or earlier when including shifted years. For longitudinal studies please remember to shift dates within one subject by the same number of days to maintain the interval information. Example: 1867-06-15T13:45:30

This could be elevated to an error when validating for public sharing.

[yarikoptic]

it might be worth to have some way to specify mode for the validator (e.g. --public) which would then check for the "originalish" dates being present.

Then cmdline etc tools used to upload could trigger that mode

[yarikoptic]

Probably actually make it --private or --policy (public|private) so by default it would raise an issue but would still allow to explicitly declare that the dataset is for internal use and original dates are ok (could be dealt with later before sharing)