Finesse resolution of inter- versus intra-file moves.

[sageserpent-open]

This follows on from #32.

That delivers a resolution where if an *intra-*file move on one side of a merge lands as an insertion that can be migrated by a single anchor or two bracketing anchors, then that final migration destination suppresses the *inter-*file move (the intra-file one is removed anyway by virtue of being migrated).

The problem is that the suppression is done in a blanket fashion, regardless of where the migration ends up. If the migration simply wonders around in the same file, we should leave the inter-file move is a divergence against the final resting place of the migrated insertion.

Similarly, only inter-file moves that land in the same file as the migrated insertion (for the equivalent sections, of course) should be suppressed.

The job here is to be more subtle about suppression.