BARUMERLI2023_METRICS - extract localization metrics

Usage:

metrics = barumerli2023_metrics(m, 'middle_metrics')
metrics = barumerli2023_metrics(m, metric)

Input parameters:

`m`	(matrix) table organized as in `localizationerror` with actual and predicted directions. The model barumerli2023 can output directly such matrix.
`metric`	(string) string indicating which metric has to be computed.

Output parameters:

varagout

output as in localizationerror-m. If 'middle_metrics' is provided then a struct will be provided.

Description:

barumerli2023_metrics(...) returns psychoacoustic performance parameters for experimental response patterns. This script is a wrapper for localizationerror. This function is also used internally in barumerli2023 which behavior is not here described. For a complete list of supported metrics, please consider localizationerror. Moreover, if 'middle_metrics' is provided the function returns a struct containing the four metrics used in the paper Middlebrooks (1999). There are: accuracy and root mean squared error for both the lateral and polar dimensions and the quadrant error.

References:

P. Majdak, M. J. Goupell, and B. Laback. 3-D localization of virtual sound sources: Effects of visual environment, pointing method and training. Atten Percept Psycho, 72:454--469, 2010.