Estimation of parameters in DDEs fails with `ERROR: TrackedArrays do not support setindex!`

[sciml]

Estimation of parameters in DDEs fails with `ERROR: TrackedArrays do not support setindex!`

[ChrisRackauckas]

If you change to the ForwardDiff backend it works:

https://github.com/TuringLang/Turing.jl/wiki/Get-started#changing-default-settings

> Turing.setadbackend(:forward_diff)
:forward_diff

> turing_result = turing_inference(prob, MethodOfSteps(RK4()), t_obs, data, [Gamma(1.0, 1.0)]; num_samples=500)
[Turing] looking for good initial eps...
[Turing.NUTS] found initial ϵ: 0.05
[NUTS] Sampling...  0%  ETA: 2:00:49[Turing.WARNING]: Numerical error has been f[NUTS] Sampling... 23%  ETA: 0:01:56

etc. We should try to get ReverseDiff working though. It works with the DiffEq solver though, so I'm not sure where it's running into an issue.

Pinging @xukai92

[ChrisRackauckas]

With the newest Turing (master) this issue with Tracked variables seems to show up everywhere. @xukai92 can you take a look at our current code and let us know what areas are supposed to be differentiated? Since when the tracked variables hit the posterior distributions they give an error.

[Vaibhavdixit02]

This should work fine now, let us know if you still face issues with the latest release.