Can we choose to load the database just once and share the memory between runs?

[d-jch]

Since load large database is time consuming, we can do it like STAR's --genomeLoad option

[Slamoth]

#719 (comment)