Add vignette on tuning multiple similar parameters for autoplot uniqueness

[davidrsch]

This pull request addresses a common pitfall when tuning Keras models with kerasnip and visualizing results using ggplot2::autoplot().

Problem:
When a Keras model specification includes multiple parameters that map to the same underlying dials parameter type (e.g., units in different layer_dense blocks), autoplot() can fail. This occurs because autoplot() cannot distinguish between these parameters without unique identifiers, leading to errors such as "Names must be unique" due to duplicated column names like "# Hidden Units".

Solution:

1. New Vignette: Introduces a new vignette, articles/autoplot_uniqueness.Rmd, which clearly explains the problem and demonstrates the solution.
2. Unique ids for tune(): The vignette emphasizes the importance of providing unique ids to the tune() function for each similar parameter (e.g., dense1_units = tune(id = "dense_layer_one_units")). This allows autoplot() to correctly differentiate and plot the tuning results for each distinct parameter.
3. pkgdown Integration: The _pkgdown.yml file has been updated to include the new vignette in the navigation bar under the "Tuning" section.

This change significantly improves the documentation by providing clear guidance on a crucial aspect of kerasnip usage, helping users avoid common visualization errors during model tuning.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!