Incorrect source position mapping when blockquotes are involved

[Knagis]

>>*foo
 >  >bar*

Both the paragraph and emphasis nodes (also the inner blockquote) will include the > characters from the second line in their contents. This is incorrect, but the fix would require that the source position mapping is stored as a list, instead of two integers.

The solution would probably require tracking of positions in each block that are taken by the prefixes so that for any child blocks or inlines these positions could be subtracted from the single position/length pair dividing it into multiple.

For now the intention is to leave it as is and revisit this issue if there are enough actual use cases for it.