ei_est() produces estimates outside [0,1] for small groups — no warning or documentation

[ashishtiwari03]

Running the README example, the vap_other estimates are well
outside [0, 1]:

vap_other pres_dem_hum 2.92 (1.46, 4.38)
vap_other pres_rep_nix -4.75 (-6.69, -2.80)

I understand that the DML estimator doesn't constrain outputs
to [0,1] by design — but this surprised me as a new user since
the estimates look like proportions.

A couple of questions:

1. Is this expected when a group has very small sample size
   (vap_other = "other" race, likely very few counties)?
2. Is there a way to get bounded estimates, or at least
   partial identification bounds via ei_bounds()?
3. Would it make sense to add a note to the docs or a
   warning when estimates fall outside [0,1]?

Happy to help draft documentation if the behavior is
intentional.