fix: don't rely on RAT internals in EnsembleSolution indexing

[AayushSabharwal]

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.

[ChrisRackauckas]

What does this fix?

[AayushSabharwal]

https://github.com/SciML/ModelingToolkit.jl/actions/runs/8950984751/job/24586851845#step:6:1792

[AayushSabharwal]

SciMLBase downstream failing is pretty critical so I'm really trying to get it working. It's led to a bunch of stuff going undetected because CI failed before the tests that would detect it ran

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 41.25%. Comparing base (a0fab7a) to head (a105744).

Additional details and impacted files

```diff @@ Coverage Diff @@ ## master #687 +/- ## ========================================== + Coverage 31.79% 41.25% +9.45% ========================================== Files 55 55 Lines 4535 4543 +8 ========================================== + Hits 1442 1874 +432 + Misses 3093 2669 -424 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[ChrisRackauckas]

Sparse(AutoForwardDiff()), but how that happened to get an incompatible version is a mystery right now @gdalle @Vaibhavdixit02 @avik-pal

[gdalle]

I have no clue what RAT or EnsembleSolution are, or how sparse autodiff is even involved here, can someone brief me in case I'm the one who should help?

[AayushSabharwal]

I think Chris was referring to https://github.com/JuliaSymbolics/Symbolics.jl/pull/1129 actually, specifically this failure

[gdalle]

I answered on the other issue ;)