feat(WENO): discretizer integration for non-uniform grids

[utkuyilmaz1903]

Overview

Non-uniform WENO-5 is wired into the MethodOfLines discretizer via WENONonUniformBoundary Val callables and grid-dispatched stencil extent. Uniform grids retain the legacy weno_f_uniform path.

Architectural Wiring

• WENOScheme(): {5,0} + [nothing, nothing] → {5,5} + WENONonUniformBoundary{1,2} / {5,4} lower/upper callables.
• extent(WENOScheme(), dorder, dx::Number) == 2; dx::AbstractVector → 0 — routed through interior_map.jl via extent(ascheme, dorder, s.dxs[x]).
• scheme_types.jl: 3-arg extent(scheme, dorder, dx) fallback delegates to 2-arg for grid-agnostic schemes.
• MethodOfLines.jl: include nonuniform_weno.jl before WENO.jl (forward-ref to WENONonUniformBoundary).
• WENONonUniformBoundary: dx::AbstractVector and dx::Number dispatches (FunctionalScheme contract).

Internal Breaking Changes

Before	After	Uniform impact
FunctionalScheme{5,0}, boundary nothing + extrapolation	{5,5}, Val boundary callables	None — uniform uses extent(..., Number)==2, interior weno_f_uniform
NU stencil extent always 2 (2-arg nothing count)	NU vector dx: 3-arg extent 0	Penultimate nodes enter interior map on non-uniform grids
extent(WENOScheme(), 1) == 2	extent(WENOScheme(), 1) == 0	None — discretizer passes s.dxs[x] (3-arg); 2-arg only affects callers omitting grid spacing

Standard uniform routing is isolated: s.dxs[x] is a Number on StepRangeLen grids → extent 2 unchanged.

Testing Highlights

• New test/Components/weno_boundary_integration.jl: extent dispatch, interior-map metadata (NU 0 vs uniform 2), get_f_and_taps routing to WENONonUniformBoundary{2}/{4}, @inferred boundary execution, discretize+solve smoke (inflow Dirichlet + outflow zero-flux).
• Integration: 24/24 pass; core logic weno_nonuniform_boundary.jl: 205/205 (no regression).

[ChrisRackauckas]

None of the test failures are related? It's getting really difficult to review these, and some time to get to green will be required to go much further faster.

[utkuyilmaz1903]

None of the test failures are related? It's getting really difficult to review these, and some time to get to green will be required to go much further faster.

You are absolutely right. I'm currently away from my desk, but my thought is that these failures are unrelated to the WENO changes. I will dive into the logs tonight or tomorrow to investigate these failures. I'll gladly help put together a fix or a separate PR to get the CI back to green so we can proceed smoothly.

[ChrisRackauckas]

Yeah let's merge but we should take a bit of time to green it.