Your interpretation of the lifetime and endtime columns is correct. I do not think that you have to worry about the angular dependence - in practice it works very well with just a distance dependent cutoff on the H-Acceptor distance (see J. Am. Chem. Soc., 1992. 114: p. 4028-4035). It may be that more distorted hydrogen bonds do actually occur in implicit solvent simulations (I have not looked at this, so it is just a speculation).
Note that with a 100 step resolution you may underestimate the hydrogen bond lifetimes; we normally use a resolution of 1-5 ps.

Lennart Nilsson
Karolinska Institutet
Stockholm, Sweden