Grammatically, the conjoined did near the end of this passage does not agree with the preceding four occurrences of thou didst. One could consider grammatically editing the last did to didst in the standard text. All five occurrences of did(st) in this passage are extant in the original manuscript. The third is spelled dids, which suggests that the conjoined did at the end might just be a scribal error for didst. Even so, there is evidence elsewhere in the text that verbs can initially have the second person singular ending, then lack that ending for later conjoined instances of verbs in the passage:
Notice that the passage ends with and will, somewhat parallel to the instance of and did here in Alma 39:3 (both lack the thou found previously in the passage). The tendency in the history of the text has been to add the appropriate second person singular ending in Alma 22:16 except for the last case, and will, the only one that lacks the thou (see the discussion under Alma 22:16). Similarly, there has been no tendency to emend and did in Alma 39:3 to and didst. The critical text will, as expected, leave unchanged the occurrence of and did in Alma 39:3 (as well as the and will in Alma 22:16).
Summary: Maintain the nonstandard use of conjoined did in Alma 39:3 (“for thou didst forsake the ministry and did go over into the land of Siron”); similar usage can be found elsewhere in the original text.