Linear model fit for data with X and Y errors using York 2004 method
YorkFit[{{x,y},...}, \"Errors\"-> err] York linear regression fit:
http://aapt.scitation.org/doi/pdf/10.1119/1.1632486
"Errors" can be supplied with:
Automatic - assumes 5% error variation in X and Y
None - no errors in x and y
p_NumberQ - assumes the erros are p fraction of the data
{pX_NumberQ, pY_NumberQ} - assumes the erros are {pX,pY} fraction of the data
{errY1,errY2,...} - errors only in Y
{{errX1,errY1},...} - errors in x and y
Weights -> None
Allows to manually specify weight matrix
"ErrorCorrelations" -> None
Correlations between X and Y errors
IncludeConstantBasis -> True
if False, this leads to fit y=b*x
achieved by adding {0,0} point with very high weights
Output: <|"fit" -> best fit, "a" -> offset, "b" -> slope, "\[Sigma]a" -> error of offset,
"\[Sigma]b" -> error of slope |>"