Enabling optimizing FFTs for dprime

[hneth]

This PR mostly enables optimizing for dprime (as goal.threshold, goal.chase, and goal). Changes include:

• computing dprime values in add_stats() helper (with correction)
• include dprime values in cue level statistics (in x$cues$thresholds and x$cues$stats)
• minor cleanups in fftrees_cuerank(), fftrees_threshold_factor_grid() and fftrees_threshold_numeric_grid()

This mostly resolves Issue #93, but the new functionality is not thoroughly tested yet. Initial tests suggest that optimizing for dprime can substantially lower other accuracy measures.

[hneth]

I noted the errors on the 3 ubuntu platforms, but the generic errors

• Error: Failed to install pandoc
• Error: Failed to get R 4.2.2

seem not caused by the changes of this PR. Will proceed to merge and hope this gets resolved somehow...