Standardize Backtest package, documentation, and release evidence #87

Audit every tracked public text file and correct confirmed identity, naming, URL, support, and claim
drift. Synchronize the factual Backtest responsibility statement across pyproject.toml, README.md,
and GitHub repository metadata. Apply the canonical author and maintainer identities, canonical
project URLs, supported Python versions, useful keywords and classifiers, README links, support
boundaries, and development gates.

Update documentation generation and deployment so strict builds record the Backtest key, package
version, and exact source commit in rendered HTML. Preserve the canonical Backtest route and add an
executable check for those fields.

Align CI, compatibility, dependency review, security, artifact qualification, release recovery,
trusted publishing, documentation deployment, tag and GitHub-release ordering, and post-publication
identity verification with the ecosystem standards. Preserve all stronger parity, licensed
VectorBT Pro, LEAN, performance, coverage, and native-behavior checks. Build the wheel and sdist once
per qualified commit, verify their declared contents, and record their versions and SHA-256 digests
in a commit-bound manifest consumed by later publication jobs.

Create a focused branch and pull request that closes this issue. Do not merge the pull request,
publish version 0.1.6, create its GitHub release, or deploy production documentation without the
corresponding explicit user approval. Apply public AGENTS.md and CLAUDE.md changes only after the
instruction audit presents exact replacements and the user approves them.

Verification must fail on a deliberately invalid input where practical. Run:

• uv run ruff check src/ tests/ plus every release-critical path
• uv run ruff format --check src/ tests/ plus every release-critical path
• uv run ty check
• uv run pytest
• pre-commit run --all-files
• uv build twice followed by reproducibility and artifact-content checks
• clean-environment wheel installation and import of ml4t.backtest
• execution of the README quick start against the installed wheel
• uv run mkdocs build --strict followed by rendered metadata assertions
• workflow syntax, security policy, dependency, compatibility, and release-recovery checks
• the ecosystem Backtest qualification against current PyPI, GitHub, and deployed-docs evidence

The issue is done when the public pull request is green, its diff contains no unrelated changes,
the private sidecar satisfies its required topology without exposing private content, and every
remaining failure is attributable only to a documented approval boundary for merge, release, or
production deployment.

Plan reference: /home/stefan/ml4t/libraries/ml4t-backtest-dev/.workspace/work/2026-09-08-ecosystem-standardization/plan.md - Issue 1 in milestone 0.1.6 - Ecosystem standardization.