[jeromedockes]

• remove old single-horizon prediction notebook which is not referenced from the book TOC anymore
• remove parallel coord json files not used anymore
• simplify pixi.tom to have just 1 environment to avoid pixi prompting when we execute tasks

[ogrisel]

Unfortunately we get a dead kernel on the prediction intervals notebooks. Maybe it's using too much memory for this runner.

[jeromedockes]

that's quite possible because it passed on the windows runner that has more memory AFAICT (and on my machine, which also has more memory)

[ogrisel]

We plan to optimize memory usage on TabICL and that might help. In the mean time let's merge. We might want to find a way to skip this notebook on macOS.

[jeromedockes]

ok let us see if this helps: #14

for the CI I will also prioritize storing a optuna db in the repo so that the random search notebook runs faster