Generalize agent_simulation for non-RA types

[maxkw]

The current version of agent_simulation calls RA_matchup_matrix. Generalize it to call the matchup matrix instead.