Fix the solution bookkeeping for save_everystep=false (#428)

[nathanaelbosch]

With save_everystep=false, savevalues! skips all of its ProbNumDiffEq-specific saving, so sol.x_filt was left with only the initial state and sol.diffusions was left empty. Two things broke as a result:

• calibrate_solution! zips sol.pu (2 entries) against sol.x_filt (1 entry), so with a calibrated static diffusion the final sol.pu[end].Σ never got rescaled by the quasi-MLE sigma^2, and sigma^2 was not recoverable from the solution either.
• Interpolating such a solution threw a BoundsError, since the interpolation indexes into the empty sol.diffusions. This is what the WorkPrecisionSet test with a dense TestSolution was marked broken for.

pn_solution_endpoint_match_cur_integrator! now always saves the endpoint into sol.x_filt, and saves the last step's diffusion when nothing was saved during the solve. The endpoint used integ.saveiter_dense for x_filt, which stays at 1 whenever dense=false; it now uses integ.saveiter like savevalues! does.

[codecov]

Codecov Report

❌ Patch coverage is 92.30769% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 91.31%. Comparing base (4c6cf29) to head (63c813f).

Files with missing lines	Patch %	Lines
src/integrator_utils.jl	92.30%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #429      +/-   ##
==========================================
+ Coverage   91.30%   91.31%   +0.01%     
==========================================
  Files          44       44              
  Lines        2230     2235       +5     
==========================================
+ Hits         2036     2041       +5     
  Misses        194      194              

☔ View full report in Codecov by Harness.
📢 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.