Record what this cycle actually verified, and what it did not

[ivanvyd]

Found while answering the question "so it's all tested and perfect?" — it was not, and this is the
part that was wrong.

The problem

This cycle's theme was that a claim in prose drifts from reality because nothing owns it. It then
left the repository's own evidence ledger describing the previous version of the code.

• docs/compatibility.md described chunk handling as though nothing had changed.
• The README's verification table never mentioned chunked results at all — so the headline
  change of the cycle appeared in neither place a reader checks before trusting it.
• Neither recorded the local packaging verification that was actually run.

Same defect class the cycle claimed to close, wearing the same clothes.

What is now recorded rather than assumed

Chunked assembly exists and has never run live. The specific unknown is named rather than
gestured at: the chunk link resolves to /api/2.0/sql/statements/…, which is not a Genie path, and
no documentation states whether the caller's identity carries there. If it does not, the client
returns the first chunk flagged as truncated — the behaviour that preceded the change. Nothing
regresses; the improvement simply does not arrive.

The service-principal recipe is documented, not executed. It follows Databricks' published M2M
flow and this project has never run it. That is a set of instructions, not a tested path, and the
ledger now says so where someone deciding whether to rely on it will read it.

The 2026-08-05 local verification, including the one check that distinguishes a package from a
build: a consumer compiled against LakeSpeak.Genie.0.1.0.nupkg reads MaxResultRows with its
documented default, so the new public member is provably in the artifact a user installs.

Verification

Documentation only; no source changes. The docs-contract test passes at 43 cases and all relative
markdown links resolve.

This PR adds no new parseable invocations — the two lakespeak mentions it introduces are
prose inside table rows, not fenced commands. An earlier draft of this description claimed the case
count rose because of this PR; it did not. The rise from 42 to 43 came from the pack run example
added in #38, which was already merged when the count was measured.