fix(eval): refuse a context window wider than the retrieval depth

[mrsibe]

Stacked on #199 (feat/eval-sweep-dashboard). Retarget to main after the chain merges.

What does this PR do?

Refuses a context window wider than the retrieval depth, instead of reporting it as if it were measured. Found in the #192 review.

The bug

The harness retrieves candidateK passages and the context metrics look at the first contextK of them. candidateK=5, contextK=8 therefore reports on five passages while claiming eight. The dashboard showed:

candidateK	contextK	Context chars	Context P
5	5	3327	0.2133
5	8	3327	0.2133

Identical — "8 is as good as 5", apparently. It was not a measurement at all: passages 6–8 did not exist. evidencePrecisionAtK divides by the passages actually retrieved, so nothing exposed it either.

The fix

A wrong number that looks like a measurement is worse than a failure, so the invariant is enforced rather than documented.

• The harness refuses it. contextK > candidateK throws before indexing, naming both numbers:
  contextK (8) cannot exceed candidateK (5): the harness retrieves candidateK passages, so a wider context window can never be filled.
• The sweep skips those cells and says so. The report lists them in a "Skipped cells" section rather than quietly omitting rows, because "we did not measure this" and "this measured the same as its neighbour" are different statements — and the old output showed the second while meaning the first.

docs/eval/sweep-v1.6.* is regenerated: 22 rows instead of 24, and the misleading candidateK=5 / contextK=8 row is gone rather than silently equal to 5.

The production cells (candidateK=20, contextK ∈ {3,5,8}) are unaffected.

Testing

• npm run typecheck — clean
• npm test — 502 pass, including 3 new harness-invariant tests that need no vector store, because the check runs before any database work: the refusal, the message naming both numbers, and contextK == candidateK being accepted
• electron . --eval-harness --eval-candidate-k=5 --eval-context-k=8 — refused with the message above
• npm run eval:sweep — 22 rows, 2 skipped and listed

Related

Part of #192 (review follow-up).