Watch this Author's files
I think your calculation of the CR is off (standard deviation is usually corrected by N-1 when using the sample-population. Cf. http://www.medcalc.org/manual/blandaltman.php)
Standard deviation lines are incorrect if the mean of the data is negative.
You didn't provide mean_nan.m and std_nan.m needed by bland_altman.m. Can you please also post them? Thanks.