ci: route CI through SciML/.github reusable workflows (@v1)

[ChrisRackauckas-Claude]

Centralize CI via SciML/.github reusable workflows (@v1)

This routes BoundaryValueDiffEq.jl's bespoke GitHub Actions workflows through the shared reusable workflows in SciML/.github, preserving the repo's existing matrices, triggers (master, paths/paths-ignore), concurrency, downstream package list, downgrade skip list, and the disabled state of the sublibrary downgrade job.

Converted workflows

File	Was	Now (SciML/.github/.github/workflows/...@v1)
SublibraryCI.yml (new)	6 per-sublibrary CI_BoundaryValueDiffEq*.yml (Core, Ascher, FIRK, MIRK, MIRKN, Shooting)	sublibrary-tests.yml@v1 — auto-discovers lib/*, replaces all six
DowngradeSublibraries.yml	bespoke per-lib downgrade matrix (disabled if: false, #417)	sublibrary-downgrade.yml@v1 (kept disabled; julia-version: 1.11, skip: Pkg,TOML, allow-reresolve: false)
Downgrade.yml	bespoke main-package downgrade	downgrade.yml@v1 (julia-version: 1.11, skip: Pkg,TOML, allow-reresolve: false)
Downstream.yml	bespoke IntegrationTest (ModelingToolkit.jl)	downstream.yml@v1 (owner: SciML, repo: ModelingToolkit.jl, group: All, julia-version: 1.11)
Documentation.yml	bespoke docs build	documentation.yml@v1 (secrets: inherit)
FormatCheck.yml	bespoke Runic via fredrikekre/runic-action	runic.yml@v1
SpellCheck.yml	bespoke crate-ci/typos	spellcheck.yml@v1
BenchmarkPR.yml	bespoke AirspeedVelocity benchpkg flow	benchmark.yml@v1 (julia-version: 1.11)

Consolidation of per-sublibrary CI

The six CI_BoundaryValueDiffEq*.yml files each dev-ed lib/BoundaryValueDiffEqCore (and ran the sublibrary's own test project) on 1.11 + lts. These are replaced by one SublibraryCI.yml calling sublibrary-tests.yml@v1, which auto-discovers every lib/<name>/Project.toml, computes the affected-sublibrary set from the internal dependency graph, and develops local path-source deps for Julia < 1.11.

Note: sublibrary-tests.yml@v1 reads per-sublibrary group/version config from lib/<name>/test/test_groups.toml (defaulting to Core/QA groups if absent). This repo does not yet have those files, so each sublibrary will run with the default group matrix. Adding test_groups.toml to each lib/<name>/test/ is a follow-up the maintainer can do to restore custom per-lib groups (e.g. FIRK's EXPANDED/NESTED).

Intentionally left bespoke

• CI_BoundaryValueDiffEq.yml (main-package tests) — runs on 1.11 and lts (Julia 1.10) and explicitly Pkg.develops all six lib/* path-sources before Pkg.test(). [sources] auto-development is a Julia ≥ 1.11 feature, so on the lts matrix entry the sublibs must be dev-ed manually. The reusable tests.yml@v1 has no hook to develop path-source deps, so converting it would break the lts job. Left as-is to avoid dropping coverage.
• QA.yml — runs test/qa/runtests.jl against a dedicated test/qa project (JET/Aqua-style QA harness). No reusable expresses this.
• TagBot.yml — TagBot automation (main package + per-subpackage subdir); not CI, left local per convention.

⚠️ Required-status-check NAMES change

The reusable workflows emit different job names than the old bespoke jobs, so branch protection required-status-check names must be updated after merge. Notable changes:

• The six CI (BoundaryValueDiffEq<Sub>) checks are replaced by the sublibrary-tests.yml dynamic-matrix jobs under Sublibrary CI.
• Downgrade / IntegrationTest / Documentation / format-check / Spell Check / benchmark job names now come from the reusable workflows (e.g. "Downgrade Tests", "Downstream Tests - All", "Build and Deploy Documentation", "Runic Format Check", "Spell Check with Typos", "Benchmarks").

Please update branch protection to match the new check names once this is reviewed.

---

Please ignore until reviewed by @ChrisRackauckas.

[ChrisRackauckas-Claude]

Follow-up fix for a coverage drop introduced by this PR: #496. The consolidated SublibraryCI.yml (sublibrary-tests.yml@v1) made BoundaryValueDiffEqFIRK run ZERO test items, because its runtests.jl dispatches on GROUP == EXPANDED|NESTED while the reusable sends GROUP=<pkg> / <pkg>_QA (no test_groups.toml present to remap). #496 restores the six bespoke per-sublibrary CI_*.yml (and DowngradeSublibraries.yml) verbatim, restores the julia-version: '1.10' pin on the Documentation caller, and keeps the faithful main-package reusable conversions. Please ignore until reviewed by @ChrisRackauckas.

[github-actions]

Benchmark Results

Click to check benchmark results

	master	174f745...	master / 174f745...
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK2()	0.638 ± 0.033 s	0.601 ± 0.011 s	1.06 ± 0.058
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK3()	12.5 ± 1 ms	11.8 ± 1.2 ms	1.06 ± 0.14
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK4()	2.65 ± 0.52 ms	2.52 ± 0.62 ms	1.05 ± 0.33
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK5()	3.1 ± 0.67 ms	3.05 ± 0.66 ms	1.02 ± 0.31
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK6()	1.47 ± 0.39 ms	1.47 ± 0.35 ms	1 ± 0.36
Simple Pendulum/IIP/MultipleShooting(10, Tsit5; grid_coarsening = false)	1.98 ± 0.71 ms	1.46 ± 0.64 ms	1.36 ± 0.77
Simple Pendulum/IIP/MultipleShooting(10, Tsit5; grid_coarsening = true)	2.95 ± 1.2 ms	2.88 ± 1.1 ms	1.02 ± 0.58
Simple Pendulum/IIP/MultipleShooting(100, Tsit5; grid_coarsening = false)	0.0386 ± 0.011 s	0.0387 ± 0.0049 s	0.997 ± 0.32
Simple Pendulum/IIP/MultipleShooting(100, Tsit5; grid_coarsening = true)	0.0601 ± 0.0085 s	0.061 ± 0.0053 s	0.987 ± 0.16
Simple Pendulum/IIP/Shooting(Tsit5())	0.232 ± 0.09 ms	0.223 ± 0.089 ms	1.04 ± 0.58
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK2()	0.801 ± 0.057 s	0.712 ± 0.0096 s	1.13 ± 0.082
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK3()	15.4 ± 5.2 ms	14.6 ± 4.5 ms	1.05 ± 0.48
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK4()	3.1 ± 0.28 ms	3.04 ± 0.18 ms	1.02 ± 0.11
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK5()	3.75 ± 0.31 ms	3.72 ± 0.3 ms	1.01 ± 0.11
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK6()	1.77 ± 0.27 ms	1.74 ± 0.18 ms	1.02 ± 0.19
Simple Pendulum/OOP/MultipleShooting(10, Tsit5; grid_coarsening = false)	3.3 ± 3.5 ms	3.29 ± 3.3 ms	1 ± 1.5
Simple Pendulum/OOP/MultipleShooting(10, Tsit5; grid_coarsening = true)	6.31 ± 6.5 ms	6.21 ± 6.4 ms	1.02 ± 1.5
Simple Pendulum/OOP/MultipleShooting(100, Tsit5; grid_coarsening = false)	0.0876 ± 0.0056 s	0.0825 ± 0.0034 s	1.06 ± 0.081
Simple Pendulum/OOP/MultipleShooting(100, Tsit5; grid_coarsening = true)	0.141 ± 0.04 s	0.129 ± 0.0041 s	1.09 ± 0.31
Simple Pendulum/OOP/Shooting(Tsit5())	0.572 ± 0.12 ms	0.575 ± 0.03 ms	0.994 ± 0.21
time_to_load	6.77 ± 0.076 s	6.67 ± 0.6 s	1.01 ± 0.092

### Benchmark Plots A plot of the benchmark results has been uploaded as an artifact to the workflow run for this PR. Go to "Actions"->"Benchmark a pull request"->[the most recent run]->"Artifacts" (at the bottom).