Upgrading from 0.4.4 to 0.5.0 about linregress

[n1m3]

I'm assuming it's this code from the substrate crate.

This appears to be a regression I introduced here.
The code in substrate forces the intercept value to 0, which results in a standard error of exactly 0, which then leads to a division by zero when calculating the t-value for the intercept.

Previously the resulting Nan was simply propagated into the p-value, but now gets rejected by the new students_t_cdf and fitting the model fails with an error.

fit_low_level_regression_model expects that 1 is passed as the intercept value and not 0, so arguably the way it is used in substrate is wrong.

However it's definitely a bug that the model suddenly fails to fit when it previously would.
I have fixed this bug by preventing the division by zero, which results in a p-value of 1.0 for the intercept in this specific case.

This fix is included in the new version 0.5.1 and upgrading to that instead of 0.5 should fix the test failures.

Please let me know if this does indeed fix the problem or if there is something I have missed.

[sam0x17]

thanks so much, that worked!