Understanding and quantifying uncertainty due to multiple biases in meta-analyses of observational studies