Try out vmap/jit in pre_slide_vmap

We get an error 137 (probably excessive memory usage?). So I will try to use jit in pre_slide_vmap function.

• Check on Converging PC
• Change add to "to_batch" suggested by Kevin
• detach pickle / dill from the init_vec and modify model parameters
• test init_vec
• goal: having an executable solution to merge in main