Regenerate Manifest.toml with Julia 1.9.2

[devmotion]

Update of #379, let's see how many tutorials are (not) broken.

Broken/assertions fail:

• Logistic Regression: Fixed by c424260
• Linear regression: Fixed by 2d64593
• Variational inference: Fixed by ef59c15
• Gaussian processes: Disabled in Transfer all high level docs from Turing.jl #375, so does not matter 😛
• Seasonal time series: Fixed by 33892a8

Fixes #378.

[devmotion]

@torfjelde The PR is almost ready but I won't be able to finish it today. It seems the VI tutorial has to be updated due to the recent changes in Bijectors - if you want, you can have a look what's the issue and push a fix. Otherwise I'll check it tomorrow.

The time series tutorial I have to run locally to see whether it's an actual problem or we just have to adjust the assertion.

Generally, in case JuliaArrays/FillArrays.jl#273 is merged and released soon, it would be nice to wait for it to avoid having to set FillArrays to 1.0.0 in the tutorials explicitly (otherwise Turing will be downgraded if you try to update the Manifest files). This would also allow us to benefit from some fixes and performance improvements in more recent FillArrays versions.

[torfjelde]

Awesome! I'll have a look at the Bijectors stuff 👍

[devmotion]

@torfjelde I fixed the VI tutorial, seems to run fine now.

[devmotion]

Once you are fine with the changes, @yebai and @torfjelde, we have to update the buildkite hashes before merging the PR (currently I have to unblock manually every buildkite run). Actually I can't do that but I'll ping Elliot.

[torfjelde]

Awesome stuff @devmotion ❤️

[devmotion]

@staticfloat can you please regenerate the keys? I switched from Julia 1.6 to 1 in this PR since the latest Turing releases don't support Julia < 1.7 🙂

[devmotion]

Thank you @staticfloat 🙂