Fix _default_dt for DelayLtiSystem (closes #544)

[baggepinnen]

Summary

• Fix _default_dt(::DelayLtiSystem) so the default time-vector dt actually adapts to the system's poles and delays. Previously the function always returned 0.07 because two […; 0] constructs collapsed r to 0, which then fell back to r = 1.0.
• Add a regression test for Small delay with step errors #544: step(delay(0.01)) now uses dt ≈ 0.00083, ran past the delay, and produces y[end] ≈ 1.
• Closes Small delay with step errors #544.

The new heuristic mirrors _default_dt(::LTISystem): take the fastest pole magnitude ω_max and the smallest positive delay τ_min, pick the more demanding of 1/(12·ω_max) and τ_min/12.

Context

Issue #544 was originally about step(delay(0.01)) failing with dt <= dtmin and a BoundsError. The error itself was resolved in #690 (2022) via force_dtmin=true, but the root-cause heuristic was never fixed — so step(delay(0.01)) ran without erroring while sampling far too coarsely to resolve the delay. This PR fixes that.

While verifying the fix I hit an apparently-unrelated upstream issue: MethodOfSteps(Tsit5()) + neutral DDE + SavingCallback errors out specifically at dt = 0.014 for the existing test system 1/(0.85s+1)·exp(-0.14s), while neighbouring dt values succeed. Reported as SciML/OrdinaryDiffEq.jl#3636. The new heuristic uses /12 (matching the LTI version) rather than /10, which happens to land on 0.012 instead of 0.014 and sidesteps the upstream bug for now.

Test plan

• delay_timeresp testset passes (33/33, including the new regression test)
• timeresp testset passes (no regressions)
• Pre-existing step(1/(0.85s+1)*exp(-0.14s), 5) test still passes (length 417, dt 0.012)
• step(delay(0.01)) now returns y[end] ≈ 1 with dt ≈ 0.00083

[JuliaControlBot]

This is an automated message.
Plots were compared to references. 4/11 images have changed, see differences below.
After pulling this PR, please update the reference images by creating a PR to ControlExamplePlots.jl here.

Difference	Reference Image	New Image
✔️ 0.0
✔️ 0.0
✔️ 0.0
✔️ 0.0

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.57%. Comparing base (2e102ad) to head (6770392).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1054      +/-   ##
==========================================
+ Coverage   91.54%   91.57%   +0.02%     
==========================================
  Files          42       42              
  Lines        5609     5624      +15     
==========================================
+ Hits         5135     5150      +15     
  Misses        474      474              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.