Incorporate estimated variance to overall, user's and producer's accuracy

[EstefaPizarro]

Describe the requested improvement
Reviewing the paper by Olofsson et al. (2014) reveals an element that could enrich the analysis of precision metrics and area estimation according to best practices (Equations 5, 6 and 7). This is the case of the estimated variance for overall, user and producer precision.

Associated sits API function
api_accuracy contains the functions for estimating accuracy and adjusted areas. Specifically, .accuracy_area_assess estimates the adjusted area and associated metrics. The code developed by FAO (https://gist.github.com/EstefaPizarro/cf00c66074581122405463c4500d5eec, from line 103) could be considered, the original repository of which is located at https://github.com/openforis/accuracy-assessment.git

In this particular script, the variance is estimated for the overall accuracy, and also for the producer and user case (considering the condition that the sampling and response design is pixel-based, according to Olofsson et al., (2014)).

Additional context
As an additional request, it would be interesting to have a visualization of the behavior of the adjusted area considering its 95% confidence interval. I leave a simple code developed (perhaps I can provide guidance so that it can be incorporated) to graph the mapped area, adjusted areas and their confidence interval for each class in a forecast estimate and area calculation.

Script to graphic confidence interval to area adjusted estimated
https://gist.github.com/EstefaPizarro/fe7813785d452a920c827f862de70a64

[gilbertocamara]

Dear @EstefaPizarro many thanks for the suggestion. We will include this improvement in version 1.5.2