Propagate interp/dense to DiffEqArrays from solution callables

[ChrisRackauckas-Claude]

Summary

• augment and all ODE solution callable methods now pass interp = sol.interp, dense = sol.dense when constructing DiffEqArray results
• Returned DiffEqArray objects are themselves callable: sol([0.0, 1.0])(0.5) works
• Adds plotting and callable interface documentation to Solutions.md, referencing RecursiveArrayTools as the canonical source for plot recipe docs
• Bumps RecursiveArrayTools compat to 3.48, 4 (requires interp/dense fields on DiffEqArray)

Context

RecursiveArrayTools v3.48+ / v4 adds interp and dense fields to DiffEqArray with a callable interface da(t) and a plot recipe that supports dense interpolation. This is the complementary SciMLBase PR that propagates interpolation data from ODE solutions into the DiffEqArray objects they return, so downstream consumers get callable, plottable arrays.

Test plan

• SciMLBase core tests pass (Pkg.test() with GROUP=Core)
• Smoke test: DiffEqArray with LinearInterpolation callable works
• Smoke test: sol([0,1])(0.5) chained interpolation works

🤖 Generated with Claude Code

[ChrisRackauckas-Claude]

Updated — now delegates full plotting to RecursiveArrayTools v4

Plotting delegation

• Moves interpret_vars, diffeq_to_arrays, solplot_vecs_and_labels, add_labels!, DEFAULT_PLOT_FUNC, plottable_indices, plot_indices, getindepsym_defaultt to RecursiveArrayTools v4
• The AbstractDiffEqArray recipe in RAT now handles the full plotting pipeline (variable selection, phase plots, dense interpolation, automatic labels)
• The AbstractTimeseriesSolution recipe becomes a thin wrapper that computes solution-specific defaults, separates discrete vs continuous variables, and delegates continuous plotting to RAT's helpers
• Net ~240 lines removed from solution_interface.jl

Interpolation propagation

• augment and all ODE solution callable methods pass interp/dense to returned DiffEqArrays
• Returned DiffEqArrays are themselves callable: sol([0.0, 1.0])(0.5) works

Compat

• RecursiveArrayTools bumped to 4 (was 3.48, 4)

Depends on SciML/RecursiveArrayTools.jl#547 (RecursiveArrayTools v4.0)