chore: docstring + naming + repr cleanup, bump 0.0.6 (closes #78)

[thomaspinder]

Summary

Wraps up the audit cleanup stack — small individual fixes that share a single style/cleanup theme.

• Set up CI with GitHub Actions (separate fast and slow test jobs) #5 volatility.py: Constant.cholesky_path / forecast_cholesky_path docstrings now reference L (the actual posterior key) instead of stale Sigma; cross-ref Constant.cholesky_at.
• Phase 4: Structural identification and analysis #8 sv/data.py: _MIN_OBS: ClassVar[int] = 24 — explicit class-attribute intent.
• Bump actions/checkout from 4 to 6 #12 fitted.py: FittedVAR.has_exog drops @computed_field; plain @property now, so it's no longer in model_dump().
• Bump codecov/codecov-action from 4 to 5 #13 sv/dynamics.py: RandomWalk / AR1 build_latent_path use "pt.TensorVariable" forward refs instead of Any.
• docs: redesign tutorial notebooks with narrative and plots #14 results.py: delete VARResultBase.__repr__ override — Pydantic's default repr is already compact thanks to Field(repr=False) on idata.
• redesign tutorial notebooks #15 src/impulso/**/*.py: docstring sweep — double backticks (RST) → single backticks (Markdown / mkdocstrings); no :meth: / :class: cross-refs anywhere.
• Svar demo #16 identified.py: delete forecast_error_variance_decomposition; fevd is the canonical name. Tutorial .qmd + .md updated.
• CLAUDE.md: codify the docstring convention.
• pyproject.toml + uv.lock: bump 0.0.5 → 0.0.6.

Closes #78. Final layer of the 4-PR audit stack, stacked on top of #77 (PR3).

Breaking changes (pre-v0.1)

• IdentifiedVAR.forecast_error_variance_decomposition removed. Migrate to IdentifiedVAR.fevd (same signature). Long name was always an alias.

Test plan

• make test — 281/281 pass.
• make check — ruff + ty clean; uv lock --locked matches pyproject.toml.
• make docs-render-ci — clean, exit 0.
• grep -r forecast_error_variance_decomposition src/ tests/ docs/ — no matches (gate 3).
• grep -rn ':meth:\|:class:' src/ — no matches (gate 4).
• grep -rn '\`' src/` — no matches (gate 5).
• SVData._MIN_OBS not in model_fields (gate 6).
• FittedVAR(...).model_dump() does not include has_exog (gate 7).
• repr(ForecastResult(...)) shows steps=4, var_names=['a', 'b'], no idata (gate 8).
• pyproject.toml shows version = "0.0.6" (gate 9).

🤖 Generated with Claude Code

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.2%. Comparing base (a24ae46) to head (6ed2c7e).

Additional details and impacted files

@@          Coverage Diff          @@
##            main     #79   +/-   ##
=====================================
  Coverage   92.2%   92.2%           
=====================================
  Files         27      27           
  Lines       1172    1167    -5     
  Branches     114     114           
=====================================
- Hits        1081    1077    -4     
+ Misses        68      67    -1     
  Partials      23      23           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.