Fix docs Euler import and Downgrade (DiffEqBase rrule(numargs) floor); Windows fixed via SciML/.github#100

[ChrisRackauckas-Claude]

Three master-CI failures addressed.

1. Documentation build (UndefVarError: Euler not defined)

OrdinaryDiffEq v7 split out the low-order RK solvers and no longer re-exports Euler from the umbrella (verified: with OrdinaryDiffEq v7.0.1, isdefined(OrdinaryDiffEq, :Euler) == false; Euler now lives in OrdinaryDiffEqLowOrderRK). The docs @example blocks in integrating.md and uncertainty_quantification.md call Euler() with only using OrdinaryDiffEq, so they errored. Add OrdinaryDiffEqLowOrderRK to docs/Project.toml and using it in both pages. Higher-order solvers (Tsit5, Rodas5P, Rosenbrock23, Vern7) are still re-exported by the umbrella and need no change. (Documentation check is green on this branch.)

2. Downgrade (Core) — _initialize_dae! MethodError in the "fail gracefully" DAE test

The first attempt (raise the OrdinaryDiffEqBDF floor to 1.7.0) did not fix it — the job still failed with OrdinaryDiffEqBDF v1.8.0 resolved, so the BDF floor was not the real cause.

Real root cause (reproduced locally on Julia 1.10 against the CI-resolved downgrade stack): DiffEqBase v6.190.2's DiffEqBaseChainRulesCoreExt defines ChainRulesCore.rrule(::typeof(numargs), f), and SciMLBase v2.146.0's SciMLBaseChainRulesCoreExt defines the same rrule for SciMLBase.numargs. With both floors resolved together the duplicate raises "Method overwriting is not permitted during Module precompilation" while loading DiffEqBaseChainRulesCoreExt; that extension fails and cascades so OrdinaryDiffEqNonlinearSolve's _initialize_dae! methods for BrownFullBasicInit are never loaded — hence the MethodError (the listed candidates were only OrdinaryDiffEqCore's CheckInit/OverrideInit/NoInit).

DiffEqBase removed rrule(numargs) from its ext in v6.211.0 (verified: present in 6.210.0, absent from 6.211.0; SciMLBase owns it now). Raising the floor to 6.211.0 drops the double-defining range. Verified locally: with DiffEqBase 6.211.0 against the CI downgrade stack the bundle precompiles with no overwriting error and the "fail gracefully" testset passes 4/4. The BDF 1.7.0 floor bump is retained (harmless, downgrade-safe) though the actual fix is the DiffEqBase floor.

3. Windows Core tests (OrdinaryDiffEqCorePolyesterExt: _polyester_batch does not exist) — NOT fixed here

The tests / Core (julia 1|lts, windows-latest) reds are not a DiffEqCallbacks bug. They are caused by a bug in the centralized SciML/.github tests.yml: the "Set test group env" step writes GROUP=Core to $GITHUB_ENV with bash syntax but no shell: bash, so on the PowerShell-default Windows runners $GITHUB_ENV is not expanded and GROUP is never set. runtests.jl then falls back to the default group ("All"), which on Windows runs the AD group (SciMLSensitivity) inside the Core test env, upgrading already-loaded packages mid-session and breaking precompilation. Linux/macOS (bash default) are unaffected.

Fix: SciML/.github#100 (adds shell: bash). Once that lands and v1 is retagged, the Windows Core jobs here go green with no change to this repo. (Tracked separately because this repo pins the reusable workflow at @v1 and does not control its contents.)

Please ignore until reviewed by @ChrisRackauckas