ci: close 2026.10.19.1 bootstrap and wheel initialization checks

[Sunrisepeak]

The released mcpp 2026.10.19.1 is mirrored, indexed and installable through exact and latest package requests. Update the workspace bootstrap pin only after those gates, and record the completed feature, release and ecosystem acceptance evidence.

The POSIX wheel smoke previously expected self env to initialize the registry. It now checks that the query leaves the registry absent, explicitly runs self init, and retains the bundled-provider assertion. The packaging documentation describes the same sequence.

Validation:

• Release 39219375110 succeeded. All 21 public assets were independently downloaded and hashed; the tag, immutable manifest and four primary/alias pairs match.
• All eight versioned GitCode resources passed full GET/SHA-256 checks before release completion; GitHub mirror hashes match the source.
• Four isolated xlings clients installed and activated exact/latest mcpp through GitHub and GitCode. Default adaptive GLOBAL selection also passed, choosing GitCode by latency. Installed binaries preserve the per-user home across package upgrades.
• A real pip-selected Linux wheel passed version, outside-venv home, read-only query, explicit initialization and copied-provider hash.
• Wheel CI 39230052671 passed actual installs on Linux x86_64/aarch64, macOS ARM64 and Windows. Publishing run 39230334718 repeated those successes; the repository environment allows deployment only from main, so its branch upload was rejected before any upload step. Main publication subsequently succeeded; see the completion record below.
• AUR publication converged with remote git/RPC and a clean Arch install. Homebrew native macOS installation/new/run and Nix complete CI succeeded.
• Four wheel-builder contract tests, version pins, documentation style/structure, generated index and whitespace checks passed locally.
• Final head 165ef5e8: main CI 39231166083 succeeded (67 successful jobs); wheel CI 39231165996 succeeded. PR rollup: 72 successful checks and two conditional skips (iOS diagnostic fallback and PR-only publication), with nothing pending or failed.
• Published fresh-install run 39228956960 succeeded in all 17 jobs, using mcpp 2026.10.19.1.
• Final multidimensional self-review was completed after all checks. No blocking findings. Runtime code remains the shipped release; bootstrap sequencing, read-only wheel queries, explicit initialization, retained-data reporting, platform behavior, upgrade continuity and documentation evidence were checked. Main-only PyPI publication and independent public wheel audit completed after merge; see below.

Refs #822, #824.

Completion record:

• Squash merge cdc8daa7c3166188a28a338ce1cb66b3e39f0719 has exactly the reviewed head's tree (c07cf38013a9e1d3f3e7a84bcf4a91758f70275e).
• Main PyPI publication 39236215439 succeeded in all six jobs, including all four actual platform installations and upload. The existing main-only deployment rule was preserved.
• Public mcpp-bin 2026.10.19.1 was independently audited: all four downloaded wheels match their public SHA-256 metadata, and both binaries in every wheel exactly match the audited GitHub release. A fresh venv installed from the public index, verified installed binary/provider hashes, queried without initialization, initialized a Unicode home, read it back in a new process and resolved it offline successfully.
• Released cold-install run 39228956960 passed all 17 jobs with version 2026.10.19.1. Both native Homebrew logs explicitly report that version and run the generated project. AUR git/RPC and a clean Arch install converged; Nix release hashes and complete CI passed.
• The temporary test-only probe PR test: collect first-run home and network design evidence #823 was closed without merge. All T1–T8 acceptance items were checked against source, actual run/PR state and retained runtime evidence; none remain open.