There are multiple ways a solution could look like. Simplest being: brc-20 indexers should only recognize inscriptions that are recognized in version 0.9 and below (and then define what that is precisely).
define a block height at which the protocol updates the ruleset to parity with the ord reference client
Whilst I’d love continued parity (recognizing all new inscription types moving forward), my concern is that this is not a one time event, and we have to run this debate back every x months/years.