Refactor plot_n_important_coeffs, unexport others

[Enchufa2]

Partial work towards #56.

So far, I've refactored the plot_n_important_coeffs just as a nn2poly plot method, and put that together with predict. We still need to

• Add an example.
• I'd like to get rid of tidytext. It's weird to have that package in suggests just for those two small things.

Please, address this in this branch while I think a little bit more about the other plot functions.

[codecov-commenter]

Codecov Report

Attention: 8 lines in your changes are missing coverage. Please review.

Comparison is base (45ca38d) 79.93% compared to head (b6eb121) 79.90%.

❗ Current head b6eb121 differs from pull request most recent head 2fcb13e. Consider uploading reports for the commit 2fcb13e to get more accurate results

Files	Patch %	Lines
R/nn2poly_methods.R	92.94%	6 Missing ⚠️
R/plot_taylor_and_activation_potentials.R	0.00%	2 Missing ⚠️

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff             @@
##           master      #68      +/-   ##
==========================================
- Coverage   79.93%   79.90%   -0.04%     
==========================================
  Files          16       16              
  Lines         648      652       +4     
==========================================
+ Hits          518      521       +3     
- Misses        130      131       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Enchufa2]

About plot_diagonal: this is just a scatter plot with an abline. It has been a useful convenience wrapper for us so that all these figures in the papers look the same. Two options here:

1. Simply unexport this function, just like with eval_poly.
2. If you think this is really important for the user, then we could integrate it into plot.nn2poly: if y is missing, then we plot the coefficients. If y is a NN object (of one of the supported classes), then we plot the diagonal.

[Enchufa2]

For plot_taylor_and_activation_potentials... I have plans. But they are quite involved, and the goal is to hit CRAN ASAP, so I'd say:

1. let's unexport both plot_diagonal and plot_taylor_and_activation_potentials for now, keeping the manual pages and using them via ::: in the vignettes;
2. let's make a first release;
3. we will decide about plot_diagonal and rework plot_taylor_and_activation_potentials in a future release.

[Enchufa2]

There's an issue with latest Posit Package Manager sync with CRAN. Tests are fine. Merging.