r2 failing for `nlme::lme()`

library(nlme)

nlme::lme(Reaction ~ 1, random = ~ 1 | Subject, data = lme4::sleepstudy) |> 
  performance::r2()
#> Warning: Can't compute random effect variances. Some variance components equal
#>   zero. Your model may suffer from singularity (see '?lme4::isSingular'
#>   and '?performance::check_singularity').
#>   Solution: Respecify random structure! You may also decrease the
#>   'tolerance' level to enforce the calculation of random effect variances.
#> Random effect variances not available. Returned R2 does not account for random effects.
#> # R2 for Mixed Models
#> 
#>   Conditional R2: NA
#>      Marginal R2: 0.000

nlme::lme(Reaction ~ Days, random = ~ 1 | Subject, data = lme4::sleepstudy) |>  
  performance::r2()
#> Warning: Can't compute random effect variances. Some variance components equal
#>   zero. Your model may suffer from singularity (see '?lme4::isSingular'
#>   and '?performance::check_singularity').
#>   Solution: Respecify random structure! You may also decrease the
#>   'tolerance' level to enforce the calculation of random effect variances.
#> Random effect variances not available. Returned R2 does not account for random effects.
#> # R2 for Mixed Models
#> 
#>   Conditional R2: NA
#>      Marginal R2: 0.486

Compare the same models to lme4 which don't fail and give different results:

lme4::lmer(Reaction ~ (1 | Subject), data = lme4::sleepstudy) |> 
  performance::r2()
#> # R2 for Mixed Models
#> 
#>   Conditional R2: 0.395
#>      Marginal R2: 0.000

lme4::lmer(Reaction ~ Days + (1 | Subject), data = lme4::sleepstudy) |> 
  performance::r2()
#> # R2 for Mixed Models
#> 
#>   Conditional R2: 0.704
#>      Marginal R2: 0.280

^{Created on 2022-06-26 by the reprex package (v2.0.1)}

Start by reproducing the two nlme::lme() examples through performance::r2(), then compare them with the corresponding lme4::lmer() results shown in the issue. Trace the performance::r2() handling for nlme mixed models and identify why random-effect variances are unavailable; done means the nlme results calculate conditional and marginal R2 without the reported warning and agree with the expected model behavior.