[FEAT] Efficient MinTrace (ols/wls_var/wls_struct/mint_cov/mint_shrink)

[elephaint]

Similar to this PR, but:

• Memory & compute efficient implementations of ols, wls_struct, wls_var.
• Efficient implementation of computing P-matrix for all MinT methods.
• Replaced eigenvalue check by Cholesky decomposition to test for positive definitiness of W-matrix in case of non-diagonal reconciliation methods, which is much faster.

[EDIT] left out the memory efficient Schafer-Strimmer implementation for now, it needs a bit more protections to avoid non-positive-definiteness.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[CLAassistant]

All committers have signed the CLA.

[jmoralez]

I think the strategy to clean the notebooks here is similar to neuralforecast:

• nbdev_clean --clear_all
• git restore nbs/examples
  Can you please try that?

[elephaint]

I think the strategy to clean the notebooks here is similar to neuralforecast:

• nbdev_clean --clear_all
• git restore nbs/examples
  Can you please try that?

Thanks - yes did so, see this PR is still from when I submitted it via a fork. I think it should be good now, and for next I'll work on the main.