See Jamison, Stephanie. 1988. ‘The quantity of the outcome of vocalized laryngeals in Indic’. In A. Bammesberger, ed., Die Laryngaltheorie und die Rekonstruktion des indogermanischen Laut- und Formensystems, 213–226. Heidelberg: Carl Winter. Jamison argued that ī is the regular development of an interconsonantal laryngeal in a final syllable, while i is the regular development elsewhere. If I'm not mistaken (I don't have a copy of the paper now) she argued that the length of ī in the weak stem of 9th class verbs was in order to distinguish the (synchronic) stem-final vowel from the (synchronic) "linking-vowel" i. That might account for its appearance in non-present forms of 9th class verbs (although once again I don't remember her argument).

In Middle Indic the outcome i seems to be more general (Prakrit gahia- for Sanskrit gr̥hīta- etc.), though I remember vaguely that early scholars took this to be a secondary (accentual) development from the Sanskrit forms. In principle they could represent the "regular" outcome of the PIE laryngeals.