Phase 2 tidy-up: revert intersphinx-qualified references once their target lectures are translated

[mmcky]

During the 2026-07-18 Track B wave, seven resynced lectures referenced lectures that exist only in the English source. Per program decision (Matt, 2026-07-18), those {doc} references were qualified with the intermediate: intersphinx prefix so they resolve to python.quantecon.org until the targets are translated. When Phase 2 lands each target lecture here, the qualified references should revert to plain local refs — a routine translate forward resync of the referencing file will do this automatically, since the source uses local refs; this issue is the checklist to make sure none are forgotten.

Referencing file	Qualified targets
lectures/ifp_advanced.md	intermediate:ifp_egm_transient_shocks, intermediate:os_egm
lectures/ifp_egm.md	intermediate:ifp_discrete, intermediate:ifp_opi, intermediate:ifp_egm_transient_shocks, intermediate:os_egm, intermediate:os_egm_jax, intermediate:os_time_iter
lectures/kalman.md	intermediate:kalman_filter_var
lectures/kalman_2.md	intermediate:kalman_filter_var
lectures/mccall_fitted_vfi.md	intermediate:mccall_model_with_sep_markov
lectures/mccall_persist_trans.md	intermediate:mccall_model_with_sep_markov
lectures/os_numerical.md	intermediate:os_time_iter

One related non-doc case: ifp_egm.md's {eq}eqvfs`` (a label defined in the untranslated ifp_discrete.md upstream) was pointed at the local, mathematically-identical eqvfs_egm instead, since `{eq}` cannot cross projects. When ifp_discrete is translated, a resync will restore the upstream form. (The upstream source referencing the remote label while the local twin sits ten lines above the sentence looks like a source-side oddity — may be worth an upstream nit.)

Grep to audit at any time: grep -rn 'intermediate:' lectures/*.md

[mmcky]

2026-07-19 audit — inventory validation, plus English-pointing refs in other series

Re-audited after today's first merge batch (8 wave PRs merged). The intermediate: table above is confirmed accurate and complete: every target resolves as a std:doc entry in the live https://python.quantecon.org/objects.inv, all 8 targets are confirmed absent from this repo (untranslated), and a scan of main plus all 69 open resync/* branches found no additional intermediate: doc refs beyond those listed.

Three further English-pointing refs exist in other series and aren't in the table. Same Phase 2 logic applies, though these revert only if/when a zh-cn edition of the target series exists, so they may be long-term residents:

Referencing file	Qualified target	Where today
lectures/ak_aiyagari.md	advanced:discrete_dp (×2)	main (branch identical)
lectures/likelihood_ratio_process.md	advanced:additive_functionals	main (branch identical)
lectures/likelihood_ratio_process_2.md	advanced:additive_functionals	main (branch identical)
lectures/wealth_dynamics.md	jax:wealth_dynamics	resync branch only (not yet merged)

All three targets validate against their live inventories (python-advanced.quantecon.org, jax.quantecon.org).

Intersphinx config health: all 11 inventories in _config.yml are reachable and serving valid objects.inv files. One nit — the tools: mapping points at https://quantecon.github.io/lecture-tools-techniques/ which now 301-redirects to https://tools-techniques.quantecon.org/. Sphinx follows the redirect so builds work today, but the mapping should move to the canonical URL when _config.yml is next touched.

Audit command covering all external series:

grep -rnE '\{(doc|ref)\}`[^`]*(intermediate|advanced|tools|pyprog|dle|dps|eqm|stats|dynam|jax):' lectures/*.md