According to ADaM IG section 2.3.2 on page 9: “The BDS is flexible in that additional rows and columns can be added to support the analyses and provide traceability, according to the rules described in Section 4.2.”
Therefor the answer for your question is “Yes, it’s allowed”.
I personally like your approach and always keep original SDTM/8601 --DTC variables in analysis files, because it’s great for 1. backward traceability and 2. for simple imputation as you described.
ADaM guideline only demostrates numeric Date/Time variable naming convention. I wonder if character Date/Time variables are also accepted. For example, TRTSDT verse TRTSDTC or TRTSDTM verse TRTSDTMC.
It’s very easy to identify TEAE by comparing AESTDTC to TRTSDTC and follow-up date after last dose in character value while AESTDTC contains partical date information. So that, imputation for missing date is not necessary.