fix(prosediff): an equal run's separator comes from whichever side has one (spec-by6.5)

spans() read Span.Space off toks[0], and for an equal run those tokens come
from the old side. An equal run exists in both revisions at once, and the two
sides can disagree about what precedes it: a block that gains words at its head
has nothing before its first old token and the insertion before its first new
one. The script then said Space=false and a renderer joining the spans wrote
"{+in practice+}the storage layer" with the words run together — a defect the
reader would read as the author's, because nothing in the output says a
separator went missing.

Not reachable from the review page today: web/diffrows.go's sideSpans already
carries a dropped span's separator onto the next kept one, and the per-line
merge never joins two spans across that boundary. It was wrong in the data all
the same, and the next consumer of the script would have inherited it.

Passing the flag into emit rather than deriving it also drops the
write-back that reached into out[len(out)-1] to clear a substitution's
separator, which would have edited the wrong span had emit ever skipped an
empty run.

feat: prosediff — word-level prose diff over markdown block structure

The Phase 0 de-risk gate. Segments a document into blocks with goldmark
(headings, paragraphs, list items, code fences, table rows, block quotes,
frontmatter), aligns the two block sequences with Myers over content
hashes, and diffs word-by-word inside modified prose blocks and
line-by-line inside modified code fences.

Whitespace and line wrapping alone produce no diff in prose, and always
do in code — that split is the whole point. Moves are detected by
verbatim anchor and grown over their neighbours, so a relocated section
does not explode into add+remove; a move that also bridges one edited
block is recognised, a section rewritten while moving is not, and that
limit is pinned by a test rather than papered over.

SPIKE.md reports the verdict against twelve real revisions of
docs/DESIGN.md: rewrapping the whole 1139-line document produces 1563
changed lines for git and zero changes here; 77% of real prose
modifications read as small edits; 13% shred and want a two-column
fallback in the web layer, which BlockChange.Similarity already gates.
Verdict: the approach works, build the review UI on it.