generated quantities block - naming conventions for posterior predictive checks and cross-validation

Clarify and standardize use of y_tilde, y_rep, and y_sim across the Stan User's Guilde.

• y_tilde is the posterior predictive on new data x_tilde
• y_rep is the posterior predictive given y
• y_sim is simulated data.
• log_lik[n] for log likelihood of data item n

Provide links to downstream analysis tools: ArviZ for Python and bayesplot, posterior, and LOO for R for posterior predictive checks and cross-validation.

Edit: added log_lik for cross-validation.

Search the Stan User's Guide for uses of y_tilde, y_rep, y_sim, and log_lik, then review the surrounding posterior predictive check and cross-validation examples. Standardize the terminology to match the issue, and add links to ArviZ, bayesplot, posterior, and LOO. Done means the guide uses the stated conventions consistently and includes the downstream analysis links.