Bland-Altman plot can be used for seeing the width of LOA and bias. Before calculating LOA/bias, we should check if the data points are scattered uniformly  along horizontal axis. Often in using chemical measurement as methodology, variability can be proportional to the magnitude of the observation (for example see the link). In such a case data should be log-transformed.

https://www.medcalc.org/manual/blandaltman.php

To the best of my knowledge, I think Bland-Altman plot is not so effective in proactively nor objectively analyzing heteroscedasticity.  Among many cases of heteroscedasticity, only magnitude-dependent variability can be detected clearly. I feel important role of Bland-Altman plot is visualization.

https://www.medcalc.org/manual/blandaltman.php

Thank you for your answer, that is very helpful. I need to use Bland-Altman plots to present data on agreement/reliability of measurement, simply agreement between two raters measuring the same phenomena by using the same tool/method. Do you mean that hetersoscedasticity can be assess only visually in Bland-Altman plots?

To the best of my knowledge, yes, hetersoscedasticity can be assess only visually in Bland-Altman plots.