Curate resource links for API reference

[shmoradims]

There are many topics that repeat throughout the API reference. For each topic, we want to have a separate link that contains more details, instead of duplicating content, or worse, stating the obvious like the following:

• Bias: The predictor's bias term.
• L2Weight: L2 regularization weight for trainer

Table below captures all the topics that we want to cover. These are the topics that apply to many APIs. For example, we have 4 or 5 cross validation extension methods, one per ML task. We can have one page explaining CV, instead of duplicating content.

Some topics already exist in glossary. We need to have a discussion about our linking strategy later. For now let's keep the list here. I suggest using the LinkPlaceholder string (tmpurl_*) in the XMLs, so that we can easily find-replace with the final link.

Topic	Description	In Glossary?	Link Placeholder	Link
Regularization	L1, L2, and other topics related to regularization	no	tmpurl_regularization	https://en.wikipedia.org/wiki/Regularization_(mathematics)
Loss	Loss functions	no	tmpurl_loss	https://en.wikipedia.org/wiki/Loss_function
Calibration / calibrators	calibrators used for producing probabilities	no	tmpurl_calib	https://en.wikipedia.org/wiki/Calibration_(statistics)
Learning rate	what is learning rate in training	no	tmpurl_lr	There's no wikipage talking about learning rate. It's always mixed with SGD or the training algorithm using it. Deleting this since it's not an independent concept.

[JRAlexander]

We can add to the glossary at any point.