Fix lcons/ucons length in __extract_lcons_ucons for optimize path

[SebastianM-C]

The per-dispatch reconstruction of the constraint length from (M, N, bcresid_prototype, f_prototype) was wrong for several discretizations:

• MIRK with prob.lcons set padded to NM (decision-var length) instead of L_bc + M(N-1)
• FIRK Expanded was off by a factor of stage+1
• The ::Nothing f_prototype fallback used N*M everywhere

The authoritative value is length(resid_prototype), which every caller already has. Replace the 3 dispatches with a single one that takes constraint_length::Int directly, and update callers.

Incidentally fixes two multiple-shooting dispatches that referenced out-of-scope bcresid_prototype / f_prototype variables.

Adds:

• Unit test in BoundaryValueDiffEqCore for both fallback and user-provided lcons/ucons paths
• Integration regression test in BoundaryValueDiffEqMIRK for the StandardBVProblem + optimize path that previously threw BoundsError

Checklist

• Appropriate tests were added
• Any code changes were done in a way that does not break public API
• All documentation related to code changes were updated
• The new code follows the
  contributor guidelines, in particular the SciML Style Guide and
  COLPRAC.
• Any new documentation only uses public API

Additional context

Add any other context about the problem here.

[github-actions]

Benchmark Results

Click to check benchmark results

	master	3e48f41...	master / 3e48f41...
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK2()	0.583 ± 0.0035 s	0.584 ± 0.0012 s	0.999 ± 0.0063
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK3()	12.2 ± 0.84 ms	12.3 ± 0.79 ms	0.992 ± 0.094
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK4()	2.77 ± 0.45 ms	2.73 ± 0.44 ms	1.01 ± 0.23
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK5()	3.25 ± 0.57 ms	3.3 ± 0.52 ms	0.983 ± 0.23
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK6()	1.59 ± 0.33 ms	1.61 ± 0.3 ms	0.986 ± 0.28
Simple Pendulum/IIP/MultipleShooting(10, Tsit5; grid_coarsening = false)	1.81 ± 0.54 ms	1.84 ± 0.53 ms	0.987 ± 0.41
Simple Pendulum/IIP/MultipleShooting(10, Tsit5; grid_coarsening = true)	3.82 ± 1 ms	3.79 ± 1 ms	1.01 ± 0.37
Simple Pendulum/IIP/MultipleShooting(100, Tsit5; grid_coarsening = false)	0.0407 ± 0.0085 s	0.0408 ± 0.0073 s	0.998 ± 0.27
Simple Pendulum/IIP/MultipleShooting(100, Tsit5; grid_coarsening = true)	0.0666 ± 0.0084 s	0.0688 ± 0.018 s	0.968 ± 0.28
Simple Pendulum/IIP/Shooting(Tsit5())	0.258 ± 0.077 ms	0.255 ± 0.075 ms	1.01 ± 0.42
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK2()	0.745 ± 0.017 s	0.745 ± 0.018 s	1 ± 0.034
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK3()	15.6 ± 6 ms	15.5 ± 6 ms	1.01 ± 0.55
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK4()	3.25 ± 0.31 ms	3.25 ± 0.24 ms	1 ± 0.12
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK5()	3.99 ± 0.48 ms	3.85 ± 0.2 ms	1.04 ± 0.14
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK6()	1.88 ± 0.43 ms	1.84 ± 0.42 ms	1.02 ± 0.33
Simple Pendulum/OOP/MultipleShooting(10, Tsit5; grid_coarsening = false)	3.71 ± 2.8 ms	3.81 ± 2.9 ms	0.973 ± 1.1
Simple Pendulum/OOP/MultipleShooting(10, Tsit5; grid_coarsening = true)	7.31 ± 5.6 ms	7.01 ± 5.4 ms	1.04 ± 1.1
Simple Pendulum/OOP/MultipleShooting(100, Tsit5; grid_coarsening = false)	0.0915 ± 0.0032 s	0.0907 ± 0.0071 s	1.01 ± 0.086
Simple Pendulum/OOP/MultipleShooting(100, Tsit5; grid_coarsening = true)	0.14 ± 0.0095 s	0.147 ± 0.014 s	0.954 ± 0.11
Simple Pendulum/OOP/Shooting(Tsit5())	0.611 ± 0.097 ms	0.63 ± 0.08 ms	0.97 ± 0.2
time_to_load	7.53 ± 0.011 s	7.28 ± 0.12 s	1.03 ± 0.017

### 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).