dicom_redact.py - does not redact all frames of all overlays when frame=overlay=-1

[howff]

When using the deid recipe rules to redact rectangles we don't know the specific frame/overlay, so we have to redact all frames and all overlays and all frames in overlays. But the code for finding out how many frames an overlay has is too complex so it's been left in the dicomimage.py library and not duplicated in dicom_redact.py, thus we only redact the first frame in each overlay. See the deduplication issue #18

[howff]

Mostly resolved with this commit 775664d

However, one remaining edge case is multiple frames within an overlay (very rare!), only the first frame in each overlay is redacted.