`bas.predict` error with a single covariate

[AleCarminati]

Describe the bug When you create a BAS linear model with a single covariate and then you try to predict new data (with se.fit=T) the function returns the following error: Error in array(STATS, dims[perm]) : 'dims' cannot be of length 0

To Reproduce

data("Hald")
hald.gprior =  bas.lm(Y ~ X2, data=Hald, alpha=13, prior="g-prior")
predict(hald.gprior, newdata=Hald, estimator="BPM", se.fit=TRUE)

Expected behavior The function should return the 95% intervals for the prediction of new data.

Desktop (please complete the following information):

• OS: Ubuntu 22.04
• R Version: 4.1.2

[merliseclyde]

Thanks! - that is a bug in the function to produce the standard errors. and affects all estimator options.

[merliseclyde]

Dev version 1.6.5 passes current tests; closing issue for now.