print.summary.cv.ncvreg behaves differently in a non-interactive session

[pbreheny]

In particular, the model summary information is duplicated

library(ncvreg)
data(Prostate)

cvfit <- cv.ncvreg(Prostate$X, Prostate$y)
summary(cvfit)
#> MCP-penalized linear regression with n=97, p=8
#> At minimum cross-validation error (lambda=0.0019):
#> -------------------------------------------------
#>   Nonzero coefficients: 8
#>   Cross-validation error (deviance): 0.55
#>   R-squared: 0.58
#>   Signal-to-noise ratio: 1.40
#>   Scale estimate (sigma): 0.741
#> MCP-penalized linear regression with n=97, p=8
#> At lambda=0.0019:
#> -------------------------------------------------
#>   Nonzero coefficients         :   8
#>   Expected nonzero coefficients:   3.88
#>   Average mfdr (8 features)    :   0.485
#> 
#>          Estimate       z      mfdr Selected
#> lcavol   0.564341  8.7910   < 1e-04        *
#> svi      0.761673  4.1676 0.0034946        *
#> lweight  0.622020  3.5220 0.0355307        *
#> age     -0.021248 -2.0908 0.6495080        *
#> lcp     -0.106051 -1.9598 0.7069159        *
#> lbph     0.096713  1.8544 0.7465772        *
#> pgg45    0.004458  1.6616 0.8050543        *
#> gleason  0.049228  0.4698 0.9362313        *