feat(fts): impact skip data for posting lists

[BubbleCal]

Stacked on #7466 (configurable posting block size).

Store per-block (freq, doc_len) impact frontiers alongside 256-doc posting
blocks, plus one level1 entry per 32 blocks, and drive block-max WAND
pruning from them instead of build-time scores that go stale as index
stats drift:

• Entries are varint-encoded for 256-doc blocks (2-3 bytes per pair vs 8);
  128-doc-block indexes keep fixed-width u32 entries.
• Bounds bake once per cached list into an Arc-shared slab (max doc weight
  per entry plus the list-wide max); per-query clones reuse the slab, so
  query time pays one multiply per bound instead of frontier rescans.
• Entry doc_up_tos decode once at construction.
• Lagging iterators park in the WAND tail under the data-driven global
  bound (query_weight x baked list max) instead of INFINITY.

Impacts shrink from 3.8 to ~0.6 bits/posting with the varint encoding.

Measured vs #7466 (its base)

Per-branch-tip wheels, 200M-doc v3-256 index (24 partitions), 1000 queries
× 8 concurrent, warm. Without impacts a 256-block index prunes on the
(valid but loose) list-level max that #7466 falls back to; the impact skip
data restores block-granular bounds on the same classic query loops:

query	#7466 (list-max fallback)	this PR
OR 3w k10	0.363s / 22 qps	0.103s / 76 qps (3.5×)
OR 3w k100	0.392s / 20 qps	0.198s / 40 qps (2.0×)
AND 3w k10	0.547s / 15 qps	0.115s / 69 qps (4.8×)
AND 3w k100	0.558s / 14 qps	0.245s / 33 qps (2.3×)

🤖 Generated with Claude Code

[github-actions]

Important

This PR touches the Lance format specification.

Substantive changes to the format specification — the .proto definitions
and the spec docs under docs/src/format/ — require a PMC vote before merge.
Minor edits such as typo fixes, wording, or formatting are excluded; use your
judgment.

If this is a meaningful format change:

• Start a vote following the Lance community voting process.
  Format specification modifications need 3 binding +1 votes (excluding the
  proposer), held on GitHub Discussions, with a minimum voting period of 1 week.
• Once the vote passes, link the completed vote in this PR. It should not be
  merged until the vote is linked.

[BubbleCal]

Rebased on the updated #7466 (which now owns all V3 breaking changes, including quantized doc-length scoring). This PR's impact bounds are baked against the same quantized lengths for the Varint (256-doc) format — quantization is monotone, so the stored (freq, doc_len) pareto frontier still dominates and the bounds stay valid upper bounds of quantized scores. No breaking change here.