ENH: Show Ns in auspice.json rather than impute Ns from attachment node

[corneliusroemer]

When a site is N, rather than showing it as N in auspice.json produced by Nextclade, we give it the base of the attachment point.

This threw me off when analyzing BA.2.75 reversion at S:493.

Implementation: This would be a matter of including mutations to N on the terminal branch, but not counting them for length, private mutations, or similar. It should be doable but requires thought and testing so it doesn't have undesired side effects.